FUNDAMENTALS OF AERIAL PHOTOGRAPHY Introduction Remote Sensing can be broadly defined as the collection of information about an object or physical phenomena without being in physical contact with the object or phenomenon. Aircraft and satellites are the most common platforms from which remote sensing observations are made. Aerial photography is the original form of remote sensing and has wide applications in topographical mapping, engineering, environmental studies and exploration for oil and minerals. In the early stages of development, aerial photographs were obtained from balloons and kites. Later, with the development of airplane (1903) and specifically during the World War (1914 to 1918), aerial photography received more atten-tion in the interest of military intelligence. In India, aerial photographs have been in use since 1920 for aerial surveys and for interpretation in specific fields such as geology. Attempts were also made to use terrestrial photographs obtained from photo-theodolites for survey purposes around 1899. Photogrammetric methods for mapping were introduced in the 1948 with the advent of multiplex stereo plotting instrument. Later, such equipment was further augmented with the acquisition of modern stereo plotting instruments during the period from 1954 to 1956. Since then, Survey of India (the national mapping agency) has kept itself abreast of the technological changes in the fields of photogrammetric mapping and aerial photography. The present discussion confines itself to aerial remote sensing only i.e. aerial photography. Aerial Photography Aerial photography is defined as the science of obtaining photographs from the air using various platforms, mostly aircraft, for studying the surface of the earth. The sun provides the source of energy (electromagnetic radiation or EMR) and the photosensitive film acts as a sensor to record the images. Variations in the gray tones of the various images in a photograph indicate different amounts of energy reflected from the objects as recorded on the film. The earth's atmosphere, which contain various particles and molecules of gases and water vapor, attenuates the incoming as well the outgoing energy/radiation (scattering) after interaction (reflectance, transmittance and absorption) with the object and thus reduces the contrast between different images formed on the photographic film. Therefore, the quality of aerial photography largely depends upon the atmospheric conditions prevailing at that time. Different filter/lens combinations can, however, be used to eliminate some of the atmospheric effects in black and white photography by making use of a yellow (minus blue) filter to reduce the effects of haze. The problem becomes more complex in the case of colour photography. Other factors that influence aerial photography are as follows.

1

Scale Scale is the ratio of distances between two images on an aerial photograph and the actual distance between the same two points/objects on the ground, in other words the ratio f/H (where f is the focal length of the camera lens and H is the flying height above the mean terrain). Due to variations in flying height, the scales of different photographs may vary. Scale may also vary because of the effects of tilt and relief displacements. Camera/Film/Filter Combinations In order to extract the maximum information from aerial photograph, the image should be of the highest quality. To ensure good image quality, modern distortion-free cameras are used. Some of the latest versions have image motion compensation devices to eliminate or reduce the effects of forward motion. Depending upon the requirements, different lens/focal length/film/filter combinations can be used. Flight Direction As a rule, aerial photography is flown in strips to cover the designated area. For convenience in handling, it is advisable to keep the number of strips to minimum. The flight direction of the strips is therefore kept along the length of the area. This direction may be any suitable direction along a natural or man-made feature and should be clearly specified. Time/Season of Photography The time of aerial photography is very important, as long, deep shadows tend to obscure details, where as small shadows tend to delineate some details effectively and are generally advantageous in improving the interpretational values of a photograph. Based on experience, aerial photography should be flown when the sun's elevation is 30 degrees above the horizon, or three hours before and after the local noontime. The choice of the most suitable season depends on factors such as seasonal variations in light reflectance, seasonal changes in the vegetation cover and seasonal changes in climatological factors. The purpose for which aerial photography is flown also dictates the season. For example, for photogrammetric mapping, geological or soil survey purposes, the ground should be as clearly visible as possible. Atmospheric Conditions As mentioned before, the presence of particles (smoke or dust) and molecules of gases in the atmosphere tends to reduce contrast because of scattering, especially by the heavier particles; therefore the best time for photography is when the sky is clear, which normally in India is from November to February. The presence of dust and smoke during the pre monsoon summer months and of clouds during the monsoon months forbids aerial photography during these periods.

2

Stereoscopic Coverage To examine the earth's surface in three dimensions, aerial photography is normally flown with a 60 % forward overlap and a 25 % side lap, to provide full coverage of the area. This is an essential requirement from the photogrammetric mapping point of view to obtain data both on planimetry and heights using the stereoscopic principle of observation in 3-D and measurement techniques with stereo plotting instruments. Stereoscopic viewing also helps in interpretation, as the model is viewed in three dimensions. Applications of Aerial Photography Mapping The application of aerial photography in photogrammetric mapping is an established procedure all over the world. It has been found to be fast, accurate, indispensable in inaccessible areas and cost effective in the long run, as initially the establishment of a photogrammetric survey/mapping unit involves capital expenditure due to the cost of photogrammetric instruments and other ancillary equipment. Interpretation Photo interpretation has revolutionalised the methods of data collection in various disciplines. It greatly reduces the fieldwork and thereby the cost. The information is reliable and acceptance for most studies such as in the fields of geology, water resources, geomorphology, hydrogeology, forestry and ecology, soil surveys, and urban and regional planning. Map Substitute In a situation where there are no adequate large-scale maps available, aerial photographs can serve as map substitutes in the form of photomaps. In the case of relatively flat terrain, these photomaps can be produced by rectification to remove the effects of tilt distortion and scale correction. This method has been found to be three to four times faster than conventional mapping by photogrammetric methods. In the case of hilly terrain, such photomaps (orthophoto maps) can be produced by the orthophoto technique, which has also proved to be faster than conventional mapping. In some urgent situations, simple mosaics prepared from aerial photographs can substitute for maps. Classification of Aerial Photography There are different criteria to classify aerial photographs depending upon the scale, tilt, coverage, film and spectral coverage/response. This classification can be defined as follows:

3

Scale * Large scale: between 1:5,000 and 1:20,000 * Medium scale: between 1:20,000 and 1:50,000 * Small scale: smaller than 1:50,000 (Scale classification may differ from country to Country) Tilt * Vertical: when the tilt is within ± 3° (nearly vertical) * Oblique : Low oblique (horizon does not appear but tilt is more than 3° ) High oblique (horizon appears) * Horizontal or terrestrial : camera axis is kept horizontal. Angular Coverage * Narrow angle : angle of coverage less than50° * Normal angle : angle of coverage of 60°. * Wide angle : angle of coverage of 90° . * Super-wide angle : angle of coverage of 120° Film * Black and white panchromatic. * Black and white infrared. * Colour * Colour infra-red/false colour Spectral Coverage/Response * Multispectral: Depending upon the number of spectral bands. As indicated above, a wide variety of photographic data products are available for mapmakers, interpreters and resources scientists from which they can derive data relevant to their specific needs. A thorough knowledge of the characteristics of these data products is therefore imperative to derive the maximum benefits and to optimize the work procedures. Photographic Products In all aerial photographic tasks, the images are recorded on film negatives, which are seldom used for mapping or interpretation. Positive prints or transparencies/diapositives prepared from the film negatives are used for photogrammetric mapping as well as for interpretation work. The criteria for good positive prints are that the prints should represent the actual response and reproduce all the details in the negative in a manner that permits easy recognition.

4

The positive materials in use are paper, film, and glass plates. Positive transparencies, which are also called diapositives, are better as, they record all the details present in the negatives. Diapositives are therefore used when high precision and quality are the goals. Paper prints are, however, so much easier to handle that they are always used for photo interpretation and field checking. The different types of photographic data products are detailed below. • Negatives on film (polyester based): previously on glass plates also. • Diapositives/transparencies on film • Contact prints on photographic paper of various grades and types. Such photographic

papers are available in grades of soft, medium and hard, and are used to obtain contact prints of optimum contrast from the original film negative. For example, if the original negative is of high contrast, a soft paper is used to prepare the contact prints. Similarly, photographic paper is also available in various thickness and surface qualities (matte or glossy) for use in different stages of mapping and interpretation.

• Enlargements obtained on film or photographic paper for specific uses. • Colour/false colour prints. Positive prints can also be prepared on colour

films/paper/transparencies from original colour negative films for use in interpretation. Likewise, such prints/transparencies can also be obtained from colour infra-red/false

• colour films. However, the processing of such films requires special processing

facilities.

• Multispectral photographs on film or photographic paper. In the case of multispectral photographs from an I2S camera, it is possible to obtain colour composites or false colour composites by the combination/superimposition of different spectral bands: for example, the blue, green and red bands can be combined in special projection instrument to obtain true colour composites. If the idea is only to view them, colour/false colour composites can be obtained from special instruments such as Mini Addcol Viewer.

Obtaining Aerial Photography As per the existing policy of the Government of India, all types of aerial photographs are classified documents (secret or restricted), depending upon the location and its strategic importance. The Surveyor General of India coordinates all activities relating to the execution of aerial photographic tasks for all civilian needs. The coordinating authority performs the following functions :

5

• Design and issue of the specifications for photographic tasks. • Layout and priorities, clearance from various agencies and distribution of tasks among

the three flying agencies. • Flight planning and evaluation for suitability of the executed tasks. • Distribution of photographs to the indenter. • Accounting for the above.

Flying Agencies As the coordinating agency does not have its own flying facilities, the flying operations for aerial photography are carried out by the Indian Air Force; the Air Survey Company, Dum Dum, Calcutta and the National Remote Sensing Agency (NRSA), Hyderabad. Cost of Aerial Photography The cost of aerial photography in India depends upon the flying agency carrying out the operation; the scale of the aerial photography; the area covered. Cost also depends whether the prints are supplied from fresh or existing photography. In the case of Indian Air Force, the cost depends upon the number of actual flying hours and the type of aircraft used: as such , the cost can not be worked out in advance. In the case of Air Survey Company, the cost is Rs. 75.20 per square mile (Rs. 29/- per square km) for 1:40,000 scale (1990 price - the cost is now under revision). For other scales, a linear conversion can be made; for example at 1:5,000 scale the cost is (40/5) X 75.2 = Rs. 601.60 per square mile and at 1:60,000 scale the cost is (40/60) X 75.20 = Rs. 50.15 per square mile. In the case of the NRSA, the cost varies from scale to scale and by the distance of the area from their headquarters. As such, the cost must be worked out separately for each task. Handling of Aerial Negatives The greatest sources of dimensional change in aerial negatives are humidity and thermal expansion/contraction. Ideally, negatives should be kept at the same temperature and relative humidity that existed at the time of exposure. The recommended relative humidity is 50 to 60 per cent, and temperatures should be 70o F with +/- 3o F tolerance. In order to ensure dimensional stability, it is advisable to control the temperature of the aerial camera while in operation so as to be close to normal room temperature. It is also recommended that negative rolls be stored for future use in controlled conditions of temperature and humidity as mentioned above. While working with negatives, their surfaces should be kept free from dust, grease, scratches and fingerprints. These precautions will help in obtaining good quality data products on reproduction as and when required.

6

Specifications of Aerial Photography For planning fresh photography, the purpose of the photography and scale are the main considerations. However, while defining these specifications, the following factors should be kept in view. Unless otherwise specified, the overlaps should be kept 60 per cent in the forward direction and 25 per cent in the lateral direction. For special tasks and terrains, the overlaps can be increased to 80 percent in the forward direction and 50 to 60 per cent in the lateral direction, especially in steep hilly areas and in city centers with high-rise buildings. • Camera lens : depending on the type of photography required. • Film/filter combination : depending on the type of photography required. • Shutter speed : depending on the scale, type of aircraft, its speed and film

speed/aperture (between 1/100 to 1/1,000 seconds). • Image motion : to be kept within tolerable limits (i.e. 20 um on the negative scale) by

the proper combination of shutter speed/aperture and speed of aircraft • Camera frame : stable mounts • Platforms : ceiling height, stability in flying and speed limits. • Auxiliary data : as required • Processing : depending on the film type and the requirements of the data products.

7

AERIAL CAMERAS Basic requirements of aerial photographs The most important uses of aerial photographs are for production of base maps and for application of photo-interpretation techniques for natural resources survey for geology, soil survey and forestry purposes. In order to be useful for above purposes, aerial photography should fulfill the following requirements : a) the photography should provide a faithful image of even the minute’s detail, b) the definition of photography should be clear, c) the photography should be distortion free and continuous, d) the tilt and crabs are within tolerable limits. Optical aspects of aerial camera In aerial survey owing to the movements of the camera relative to the ground, short exposure time and the necessity for bright photography, the aerial camera should fulfill the following requirements : a) A large relative aperture of the taking lens to produce bright and clear photographs. b) The photographs produced are geometrically accurate with a high degree of sharpness

and good definition over large angular field. c) The camera lens should be free from following lens aberrations. I. Spherical aberration - Occurs when rays from various zones of a lens focus at

different places along the axis; this results in an object point being imaged as a blurred circle. It is caused by the spherical shape of the lens surfaces. It is decreased as the lens aperture is reduced.(Fig.1).

II. Coma - is a comet-shaped blur of light formed around image points off the axis. It

is partly due to spherical aberration of oblique rays. III. Astigmatism - is an aberration, which causes a point object off the axis to be

imaged as two mutually perpendicular short lines, located at different distances from the lens. One of these is radial and other tangential with respect to centre of the field.(Fig.2).

8

IV. Curvature of the field - The surface of the best definition is located midway between the two radial and tangential surfaces as explained in (iii) above and its departure from flatness is termed `curvature of the field.' (Fig.2).

V. Chromatic aberration - results when rays of various wavelengths of different

colours focus at different distances from the lens. Lateral chromatic aberration is a difference in image magnification for various colours caused by chromatic aberration of oblique rays (Fig.3).

d) The camera lens is free from lens distortion. Radial lens distortion is the linear

displacement of an image point radially to or from the centre of the image field - a positive value being considered away from the centre. Tangential lens distortion is a small displacement in the image plane perpendicular to radial lines from the centre of the field and is caused due to either lack of precision in centering of the various lens elements or to improper mounting of the lenses. A lens exhibiting distortion will image a square positioned perpendicularly and symmetrically with reference to the optical axis as a pincushion or barrel since the various zones of the image correspond to different focal length values and consequently varying image scale (Fig. 4).

e) The definition is good. Definition concerns the ability of a lens to record fine details and

can be expressed as maximum number of lines pair per millimeter that can just be seen as separate lines in the image. Normally, a resolving power of 45 lines pair per mm is considered satisfactory.

Aerial cameras The aerial cameras should be of a good quality. Its optical unit holding the lens, fiducial marks and edges, which define the focal plane, should be of a rigid mechanical structure. The main types of aerial cameras are given in Table 1. Components of Aerial Cameras The major components of an aerial camera are : Lens, lens cone, shutter and diaphragm, camera body, drive mechanism, film magazine, focal plane and film flattening device (Fig.5.). The lens should be distortion-free and of high resolution. The lens surfaces should have anti-reflection coatings. The lens cone support the lens and retain it at a predetermined distance and position from the film or plate negative, and serves to include direct light from striking the

9

film or plate. The interior of the lens cone should be black and fitted with baffles so as to reduce the reflection of flare light. The shutter and diaphragm of an aerial camera functions as a light value and regulates the amount and period of time that light is permitted to pass through the lens and expose the film or plate. The shutter should be of the between the lens type.

Fig. 1 Spherical aberration Fig. 2 Astigmation ( Dotted curves) and curvature of field ( solid curve )

Fig. 3 Chromatic aberration Fig. 4 Effect of lens distortion

10

Fig. 5 An aerial camera

( a= film, b= pressure plate, c= focal plane frame, d= lens, e= filter)

11

The camera body houses the camera drive mechanism, driving motor, operating handles and levers, electrical connections and switches and other accessories which may be necessiciated by specified requirements. The camera derive mechanism is the power unit and power distributor for the entire camera. The electric motor causes the many cams gear and shafts of the camera to move. By means of rods and couplings, the power is routed to the shutter and the film magazines. When a cycle is completed, the camera drive receives and electrical or mechanical impulse, operates the shutter, and thus exposes the sensitized material. The film magazine is first of all a container of film. Besides this it contains a driving mechanism, which receives power from the camera drive mechanism and thereby shifts the film after each exposure has been made. In addition, the magazine contains a means of holding the film flat in the focal plane while the exposure is being made. The focal plane of an aerial camera is the plane in which all light rays through the lens cone come to a focus. A frame bound the focal plane, which determine the size of the negative. In order to provide a means for placing the emulsion of the film in the exact focal plane, a metal plate known as locating back is used in modern aerial cameras. The film flattening is usually accomplished in modern aerial cameras by a vacuum system. The locating back has grooves in which there are small holes which leads to a central vacuum connections and hold the film firmly against the focal plane frame. Camera mounting It is advisable to keep the maximum relative motions between image and film, arising from angular vibrations during the longest exposure, below a value of 0.002 mm. For a good camera mounting, the centre of support should be near the centre of gravity, the mount should be near the centre of gravity, the mount should feel soft and yield easily to hand pressure, with its natural frequency not higher than 5 cycles per second, and damping should be somewhat under damped. Intervalometer The use of an intervalometer, which controls the automatic exposure of the camera at, specified distance intervals, along the flight line result in correct forward overlap. The determined exposure interval is set on the intervalometer, which is then regulated by electric or mechanical impulses with varying flying speeds and flying heights. Crab compensation Another important requirement is that the camera must be able to be turned into its mount to compensate for crab. The crab is determined through simple sighting devices

12

and is eliminated by turning the camera ;through the `angle of crab' and thus uniform overlap over the entire breadth of the photograph is ensured.

13

TABLE I DETAILS ABOUT CAMERAS

SL.NO.

CAMERA TYPE & MAKE

TYPE OFLENS

ANGU-LAR COVE-RAGE

FOCAL LENGTH

PIC-TURE SIZE

SHUTTER TYPE &SPEED

FILM LENGTH USED PER ROLL (m)

NO. OF EXPO-SURE PER LENGTH

OVER-LAPS IN FLIGHT DIREC-TION POSSIBLE

SHORT-EST SEQUE-NCE OF PICTURE

REMARKS

1. Wild RC 5(a)

a)Normal angle b)Wide angle

60o

90o

210 115

18x18 18x18

Spring type 1/100 1/200

60 60

280 280

20,60,70,80 -do-

3..5 -do-

Lens fully corrected for visible spectrum (400 nm to 750 nm) -do-

2. WildRC 8

Wide angle 90o 115 18x18 1/300 60 280 -do- -do-

3. Wild RC9 Universalwide angle

90o 115 23x23 Rotaryshutter with continuous setting from 1/100 to 1/00

60 235 -do- -do-

4. Wild RC10

Wide angle F/5.6

90o 152 23x23 Rotaryshutter 1/100 to 1/1000

60 or 120 230 or 460

20,25,30 50,55,60 65,70,75 80,85,90

1.6

5, ZeissRMK’A

Wide angle 93o 153 23x23 Rotatingdisc type from 1/100 to 1000 continuously

120 470 20 to 90 continuously

2.0

Lens corrected for the visible and infrared sectors of spectral range (500 nm to 900 nm) and with appropriate light filters

6. MultibandI2S

76o 153 8.9x8.9 Focal plane(2 type A, B)

76

1/140 to 1/350-A 1/350 to 1/980-B

- - 2.0 Can be used for photography on Panchromatic, infrared, colour and false colour infrared films.

The older types of cameras of Eagle IX type of British make, which have been used largely in the past, are also available. These cameras have lenses, which show distortions, which are appreciable, and the image quality is also not good as with modern cameras. The focal length available is 6”, 10”, 12” and 20”.

14

PROCESSING OF BLACK AND WHITE, COLOUR, BLACK AND WHITE INFRARED AND COLOUR INFRARED FILMS, FILM DENSITY AND

CHARACTERISTICS CURVES A real image formed by an optical system, may be recorded permanently by situating a photographic film in the image plane. After the appropriate processing and printing, a two dimensional impression of the scene is obtain. BASIC BLACK & WHITE PHOTOGRAPHY A light sensitive emulsion consisting of a suspension of microscopic silver halides crystals in a gelatin binder is coated on a piece of flexible polyester to produce a black and white film. How a picture is formed 1. Light reaching the sensitive layer through the camera lens during the exposure causes

an invisible change in the silver halides. 2. When the emulsion is treated with a developer those grains of silver halides, which have

been affected by light, are reduced to black metallic silver. The developer has no effect on the grains in areas that have received no light during the exposure.

3. The grains that are not affected by the developer would blacken if exposed to light and

so a fixing solution is used to dissolve these unexposed, undeveloped grains without effecting the permanent image in black silver.

4. The film is then washed to remove all unwanted chemicals. And so a negative is made in

which the various brightness of the original are recorded as corresponding degrees of blackness.

5. To obtain a picture in which the various brightness correspond to those of the original

scene, this negative is printed by passing light through it onto a paper coated with a silver halides emulsion.

6. The paper is then developed, fixed and washed as in the case of the negative. Physical properties of the developed image The degree of darkening of the film on development is expressed by a logarithm number, which is called Density. The higher the density, the darker is the film. In case of film, we are only interested in transmission density. Following relation is used for defining photographic density.

15

Illumination (OUT) Transmittance = ------------------ X 100 percent Illumination (IN) Illumination (IN) Opacity = ----------------- Illumination (OUT) Density = Logarithm of Opacity Characteristic Curve The relation between exposure, development and the density of corresponding negative is represented by a characteristic curve. It is also known as H and D curve, D log E curve or response curve. In the characteristic curve, densities are plotted against the logarithm of the exposure to which they correspond. The characteristic curve of all photographic films or papers has general S shape as shown in the figure 6. The shape of the curve will vary on the following conditions. a) Type of emulsion (speed, contrast, B&W or colour etc.) b) Type of developer (contrast, chemical constitution etc.) c) Time of development d) Temperature of developer e) Dilution of the developer f) Method of agitation (manual or automatic processing) The lower part of the curve AB is known as the Toe region and the upper part CD is known as the shoulder region. The central part BC is known as straight-line region and tangent of its angle with the log. Exposure axis is known as Gamma or contrast of the emulsion. Straight-line region is the best region of the film response. The aim of correct exposure is to utilize the straight-line region of the characteristic curve. Speed/Sensitivity Speed is one of the most important sensitometric properties of photographic material intended for aerial photography and also for general-purpose photography. Knowledge of speed value permits the proper settings of shutter speed and aperture number for correct exposure of a film. The general sensitivity of a film describes its ability to produce a density change on exposure to light. The less light required to produce a certain measurable density response, the higher is the sensitivity/speed of the emulsion. Speed of the film is directly proportional to the size of the silver halides grains.

16

Speed of a film is specified by arithmetical numbers in American and British standards (ASA/BS) and by logarithmic numbers in German Standard (DIN). The formula to compute these speed number is :- 0.8 ASA/BS = ---- Em DIN = 10 log (1/Em) Here Em is the exposure in Lux seconds corresponding to the point M obtained on the characteristic curve for a density value of 0.1 above base plus fog density. In case of under exposure, we have loss of information in the lowlight areas we have loss of information in highlight areas for over exposure. Printing from negatives There are two methods of photo printing. a) Contact printing : The negative film and the positive printing paper is kept in perfect

contact and exposed to light without a lens. Size of the positive print is of the same size of original negative.

b) Projection printing : The negative image is projected through a lens on the positive

paper and the size of the positive print is of the desired magnification. For preparing good quality positive prints from B&W negatives of varying density range or contrast we use positive papers of different grade such as hard, medium, soft etc. BASIC COLOUR PHOTOGRAPHY In the spectrum of light (see fig.7), the most obvious colours are blue, green, red and spectral yellow. For convenience, in colour photography, the spectral yellow is ignored (it is a very narrow band of wavelengths) and it is said that the spectrum is divided into three major bands, each being one third of the total : BLUE (400 to 500 nm) GREEN (500 to 600 nm) and RED (600 to 700 nm) and that WHITE LIGHT contains equal quantities of these three. All systems of 'true' colour photography in use today are based on three facts. They are

17

1) All the colours and wavelengths of light that the human eye can see can be matched by mixtures of only three suitably choosen colours called YELLOW, MAGENTA & CYAN. Each of these absorbs one third of all the wavelengths in white light while transmitting the other two thirds. See figure 7.

Fig. 6 Characteristic curve and aerial film speed

Fig. 7 Visible spectrum with transmission and absorption characteristics of filters

18

for additive and subtractive colors. Because each colour absorbs the one third of the spectrum that it does not transmit, each is sometimes called a MINUS COLOUR. That is YELLOW is called MINUS BLUE because it absorbs Blue MAGENTA is called MINUS GREEN because it absorbs Green CYAN is called MINUS RED because it absorbs Red. The three colours BLUE, GREEN & RED, each being one third of the total spectrum are called PRIMARY COLOURS whilst each of the other colours, YELLOW, MAGENTA & CYAN are called secondary or COMPLEMENTARY COLOURS. YELLOW is complementary to BLUE (Yellow absorbs blue) MAGENTA is complementary to GREEN (Magenta absorbs green) CYAN is complementary to RED (Cyan absorbs red) The three complementary colours are used to match all of those in nature, obviously anything, which was yellow, magenta or cyan is easy to match and (figure 8) shows how blue, green and red are made. White is the absence of any colour whilst black is a mixture of the maximum possible quantity of all three, grays are matched by mixtures of equal quantities of yellow, magenta and cyan but not at the maximum possible strength. The correct proportions of each of the complementary colours can match any other colour that occurs in nature. The other two important facts, which are used in colour photography, are :

2) It is possible, in effect, to make three different emulsions, each sensitive to one third

of the spectrum and to use then all at the same time, that is one emulsion is sensitive to blue light (400 - 500 nm) another is sensitive to Green light (500 to 600 nm) whilst the last is sensitive to Red light (600 to 700 nm).

3) In the processing of the colour film or paper, it is possible to produce a different dye in

each emulsion layer and each dye may be in the form of an image complementary in colour to the sensitivity of the emulsion layer in which it is formed.

That is, the Blue sensitive layer gives a Yellow dye image. the Green sensitive layer gives a Magenta dye image. the Red sensitive layer gives a Cyan dye image.

19

This system is called CHROMOGENIC development. In respect of point 3 above, the three layers will only be sensitive to Blue, Green and Red, and the dye images produced will only be Yellow, Magenta and Cyan respectively in a 'true' colour film. In a 'false' colour film, there will not necessarily be three emulsion layers, the emulsion layers may not be sensitive to Blue, Green and Red but possibly to other bands of the visible spectrum or even to invisible radiations such as infrared, nor will the dye images produced necessarily be complementary to the sensitivities of the emulsion layers in which they are formed. Construction of a colour film The sources of all contemporary colour photographic systems depends not only on the facts that various emulsions can be made sensitive to well defined spectral bands and that dye images can be produced later in those emulsion layers by chromogenic development, but also on the facts that all the emulsion layers are exposed at the same time and that the images formed in them are exactly registered. In other words, the image of a point in one layer will be exact coincidence with the image of the same point in all the other layers. The three (usually) emulsion layers and the three images formed cannot be physically separated from each other. A `true' colour film consists, then, of three light-sensitive emulsion layers coated upon a film base, each emulsion being sensitized to one of the primary colours and capable of producing the corresponding complementary colour during processing, the dye image which is produced by chromogenic development in each of the three emulsion layer will absorb the primary colour to which that layer was originally sensitive. This system is used for both types of film, negative and reversal, giving respectively, negative and positive images, figure 9 shows the cross-section of the true colour film. ---------------------------------------------------------------------------------------- BLUE SENSITIVE EMULSION gives YELLOW DYE IMAGE yellow layer GREEN SENSITIVE EMULSION gives MAGENTA DYE IMAGE RED SENSITIVE EMULSION gives CYAN DYE IMAGE FILM BASE FILM BASE ---------------------------------------------------------------------------------------- Figure 9 - Cross-section of a true colour film In figure 9, the layer between the Blue and Green sensitive emulsion layers was not explained. It is a yellow layer which is necessary to absorb blue light and prevent it from reaching the lower two emulsions because they are also sensitive to blue light. As was

20

stated earlier, all photographic emulsions are inherently sensitive to blue and in a colour film, this sensitivity is suppressed by making sure that no blue light reaches the Green and Red sensitive emulsion layers. The yellow filter layer is destroyed during processing.

Fig. 8: Color triangle showing the relationship among additive (+) and subtractive (-) primary colors.

21

Fig. 9: Cross sections of positive and negative color films showing how images are formed on the three emulsion layers. Colour film type It has already been mentioned that there are two types of colour film, negative and reversal, giving negative and positive image respectively. The two types of film are constructed in an identical way but their emulsions are slightly different, it is the processing of the film, which controls the final image depending upon whether it is negative or positive. The best images can only be produced however, if a film is made to be processed only to a negative or is made to be processed only to a positive, it is not possible to make just one film that can be processed successfully to either a negative or a positive. This has been attempted in the past (Kodak `Ek-tachrome `MS' `Aerographic' film) but by the most modern standards, the quality achieved was inadequate. In other words, manufacturers make and sell two different types of colour films for negative or positive images, they are known as negative or reversal films. It is possible to buy `true' colour films as either negative or reversal material but the most widely used `false' colour (Kodak `Aerochrome' Infrared) is available only for reversal processing to give positive images. Of the other `false' colour films, Kodak Water Penetration Colour films is a reversal film, the G.A.F. Blue-sensitive Colour Film is also for reversal processing whilst the Russian films which are quite well known but no longer available (Spectrozonal) were made in two types, one to give negatives, the other to give positives by reversal processing. Colour reproduction The next four diagrams show fig.10 the reproduction of colour by an aerial colour negative film fig.11 the reproduction of colour by a 'true' colour reversal film fig.12 the reproduction of colour by Kodak 'Aerochrome' Infrared 'false' colour film fig.13 the reproduction of colour negative by colour printing paper. Choice of film type Colour photographs made with reversal film are usually sharper, contain more detail and have better colour reproduction, (in true colour photography). BUT, the film actually used in the aerial camera is the same film that is processed to give the final diapositive, it

22

is the same film that is exposed in the air, which is finally used, by the interpreter or the photogrammetrist. This is a poor situation because if a diapositive is damaged or destroyed, it can only be replaced by re-flying the photographic mission. It is possible to make paper prints or duplicate diapositives from original transparencies but the techniques, which must be used, are difficult and the material is expensive as

Fig. 10: The colour negative - positive process

23

24

Fig. 11: The substantive reversal process

25

Fig. 12: Reproduction of colours by Kodak Aero chrome Infrared False colour film

26

Fig. 13: The silver - dye - bleach process compared with the costs of making similar products from colour negatives - and in some areas of the world, the adequate materials may not be available. It is therefore better to take the original aerial photographs on colour negative film. Although a diapositive made from a negative will not be as good as an original diapositive but will be very similar to a duplicate diapositive. The reasons for the changes in quality are that, as with black and white techniques, there is always a slight loss of resolution and detail each time the photographic process is used but, in colour photography, each time the process is used, the colour itself also becomes worse. It is obvious, then that the user of colour aerial photography has some difficult choices to make. It is better to use colour negative film for large-scale aerial photography to avoid the overall blue cast due to effect of scattering of lower wavelengths. And for small scale aerial photography IR colour aerial films are suitable. If all photography is carried out with reversal films, both true and false colour, then the optimum quality will not be achieved and some production will be very slow or even impossible. There are no straightforward solutions to these problems. Aerial photography should be based on utilization to fulfill the desired results of mapping/interpretation.

27

Equipment for colour aerial photography

So far, nothing has been written in these notes about the characteristics of the

cameras and accessories, which should be used for colour aerial photography. The reasons for this is that there are no differences in the cameras required - any modern aerial camera with lenses suitable for both panchromatic and infrared black and white photography will also be suitable for colour work. The one big difference in the taking of colour photographs is that in true colour work, it is not permissible to use a yellow filter to reduce the effects of haze. But, because much of the haze radiation is short wavelength, it is both possible and desirable to fit the camera with an `ultraviolet absorbing' filter which transmits virtually all visible light but absorbs the unwanted ultraviolet radiation. Figure 14 shows the transmission curve of a typical filter that might be used. Another aspect of importance, in any use of colour film, for any purpose, is that the exposure of the film must be correct, it is not possible to compensate for under - or over-exposure during the processing of the film. Correct exposure is absolutely essential for colour reversal films. Should colour photography be used?

If the use of colour film is considered, it is unlikely to totally replace black and

white film but may be in addition. The image quality is still not quite as good as that of black and white photography and this is one reason why it is sometimes said to be desirable to take colour photographs at a larger scale. Haze also has a detrimental effect on `true' colour photography that cannot be minimized by the use of a `minus-blue' filter. These facts make it desirable to produce colour photography at the largest possible scale for optimum quality and information content. For reasons of cost, economy and time consumption, it may be advisable therefore, to cover the total survey area by black and white photography and then to add colour photography for those parts where this might have particular advantages. Colour is especially useful when results in black and white do not show sufficient differentiation between important details and where colour differences are clear and relevant to the investigation being carried out. The much higher costs of aerial photography are not caused just by the higher prices of colour film, these are only a very small fraction of the total expenditure on a photographic mission, they are caused much more by organizational and time consumption problems, not the least of which is the necessity of waiting for really clear atmospheric conditions. The instances in which colour aerial photography can be the best means of solving particular survey problems are limited but are becoming decreasingly so as new and better materials and methods become available. If colour photography gives better information, then the time and the money spent to produce it can be well justified.

28

AERIAL FLIGHT PLANNING

General After all the relevant information regarding the aerial photography has been obtained, the photographic flying mission has to be carefully planned. The fundamental requirement of flight planning is adequate stereoscopic coverage with the least number of pictures consistent with desired accuracy. The main factors affecting flight planning are selection of aerial camera, flight altitude, forward overlap, lateral overlap, flight plan, selection of aircraft, aerial film, and navigation instruments. On finalization of flight planning the actual flight takes place when the weather conditions is ideal and the time and season of photography is desired. Occurrence of clouds, avoidance of crabs, water surfaces and gaps are also considered. Selection of aerial camera If the aerial camera and focal length of the lens have already been specified by the inventor, then these should be used. If not, the focal length of the lens and aerial camera has to be decided by Survey of India, who is responsible for design of photographic specifications. Aerial cameras may be of different format size. For reasons of economy, the larger format size should be used as far as possible. Flight altitude The scale of aerial photographs to be flown which is to be indicated by the user is defined as the ratio of focal length(f) and the flying height(H) i.e. f/H. Thus, the photo scale over an area can be constant if in other words the terrain height is constant. Usually in nature, there is altitude variations in the terrain and therefore, the scale also vary from photo to photo or even point to point. Thus, the given photo scale is only the mean scale which is designed to be achieved by computing the flight altitude above mean ground level. The flying height above mean ground level is important as it in intimately connected with the ceiling height of the aircraft which is the maximum altitude above mean sea level at which the aircraft can fly safely . If the height of mean ground level above mean sea level is h and ceiling height of the aircraft above mean sea level CH, them maximum flying height above mean ground level H for particular aircraft is given by, H =Ch-h. However, the flying height above mean sea level which is the height of the aircraft above mean ground level added to the height of the mean ground level above mean sea level and is given by H + h is important for the pilot as he has to maintain the height of his aircraft at this level. Scale variations As explained earlier the photo scale is not constant over the entire area and the expected scale variations can be computed according to terrain height variations. However, it is also difficult to keep the flying height constant. The determined height can be held

29

within ñ50 feet while the flight is in progress. Actual flying height shall generally be within ñ2% for ñ200 feet of the computed flying height. Alternatively it is possible to maintain a near constant photo scale over an area with terrain variations by computing different flying height for each strip. The constant photo scale is not so important for photogrammetry as it is for preparation of semi-controlled photo-mosaics. If the photo scale is kept constant, scaling of individual photographs by means of photographic enlargers or rectifiers can be avoided while preparing semi-controlled mosaics. In order to obtain a constant scale, flying height is determined by means of flight altimeters that are accurately calibrated to international standard atmosphere. This calibration has to be carried out periodically on the test bench and has to be duplicated or repeated during flight. Date in Flight Report For purposes of checking the scale it is necessary to mention the most important data in the flight report, and therefore, the following should be mentioned therein. a) Altimeter calibration data. b) Sea level pressure or aerodrome level pressure of take-off. c) Sea level pressure or aerodrome level pressure at landing. d) Sea level pressure used at the survey area. e) Corrected outside air temperature (COAT) for each 1000 feet. f) Instrument’s and installation's errors calibrated for the flying height. g) Indicated or calculated flight altitude or flying height. h) Required true altitude. Forward Overlap: For stereoscopic viewing of photographs there should be certain amount of overlap between two consecutive photographs. The overlap in flight direction shall be 60% ñ 5% as expressed with respect to average terrain datum. In no case, however, less than 53% and also in no case the forward overlap between photo number 1,3,5 and etc. be less than 6% at the highest point. Other forward overlap percentages can be specified if other special requirements have to be fulfilled. In cases of block photography where saving in ground control is envisaged a fore lap of 80% to 90% is used and suitable photographs having matching edges

30

with photographs of adjacent strips, but having at least 60% overlap amongst the photographs of the same strip, are chosen for actual work. Where the end of the strips of one block overlap the end of strips of another block, the overlap shall be at lest 3 photographs and preferably 6 photographs. This is necessary to make use of the control points of the existing photography for the new photography. The above recommendations can be used for all normal cases but in cases of mountainous terrain with large relief variations, the tolerance of ñ 5% in forward overlap may not be sufficient if the highest ground points are to have a minimum overlap of 53%. Thus, it will be safer to have a slightly more forward overlap i.e. 60% to 65% in such cases. Lateral overlap The lateral overlap between strips only requires being sufficient to provide certainty of identification of common detail and to allow for the lateral tilt and slight deviations from course in the length of the strip. In general, a minimum lateral or side overlap should be aimed at, for reasons of economy. In majority of cases, an average lateral overlap 20% of the photo format size can be specified. Tolerance must be allowed at about 5% for navigational uncertainties and 5% for small terrain height differences. In terrain with relief variation not more than 5% of flying height, lateral overlap specifications may be stated as the lateral or side overlap shall be 20% ñ 10% of the photo format size. This results in maximum value of 30% and minimum value of 10% lateral overlap. The effect of relief in mountainous terrain is to cut out the effective coverage due to the scale of photography being larger on hilltops than in valleys and therefore adequate provision should be made for relief at the planning stage. Based on past experience in the Himalayas, it has been found that a lateral overlap of 35% caters for the terrain relief variations in these areas. In mountainous areas the specifications for lateral overlap may also be given as 20% + 10% + times relief percentage, where percentage relief. Extreme difference in ground height in the overlap area X= --------------------------------------------------------------------- x 100 height of aircraft above lowest ground Selection of aircraft The selection of aircraft is done by the flying agency. The two factors, which are required to be considered for the selection, are, the ceiling height of the aircraft and its flying range. An aircraft, to be suitable for aerial photography, should have requisite speed, a high rate of climb, and good stability while in flight and unobstructed view in all directions for ease of navigation. It should have a ceiling height equal to or higher than the highest-flying altitude specified. It should be able to remain in the air long enough to take advantage of suitable photographic time, roomy enough to carry its full load to the maximum flying height specified.

31

Navigation instruments and crew If any navigation instruments e.g., radar or Decca navigator, inertial navigation system, global positioning system is being used, these should be checked before installation. All the spare magazines should be checked. Aerial camera and viewfinder should also be checked for satisfactory operation. The photographic crew i.e., the pilot and aerial photographer-cum-navigator should be well qualified for the photographic task assigned to them. Special survey data In case any special survey data are required, these should be specified clearly. The special survey data may be one or more of the following : a) Statoscope reading for each camera station. b) Horizon camera images for each exposure. c) Radar altimeter of Airborne Profile Recorder. d) Aerodist data. e) Gyro-controlled or inertia-controlled camera vertically data. f) Doppler-controlled air base measurements. Aerial film A fine-grain emulsion aerial film manufactured by any of the established manufacturers, e.g.Agfa, Gaevert, Ilford or Kodak should be used. Flight instruments and aircraft's calibration According to the ICAO standards, the flight instruments shall be calibrated at least once in every 1.5 years.This applies in particular to the barometric altimeter, temperature gauge, and radar altimeter and for any other available scale or altitude reference system, the magnetic compass, and direction gyro. The calibration shall consist of individual calibration of each instrument and of a calibration of the total instruments system in flight in order to determine and to correct for installation errors and for operational performance errors.

32

BASIC GEOMETRIC CHARACTERISTICS OF AERIAL PHOTOGRAPHS Introduction Photographs taken from an aircraft commonly termed as aerial photographs have come to play an ever-increasing role in the execution of cartographic mapping on various scales and in evaluation of natural resources of a region. Uses of aerial photographs in other fields are also manifold, in fact the scope seems limitless. All studies of natural resources basically involve qualitative examination of the terrain, the correct correlation of the observed data and finally the evaluation of the data. The Orthodox method involve detailed study of the terrain with its attendant handicaps but modern techniques of investigations make full use of the immense wealth of information which is recorded on an aerial photograph and thus not only economize and expedite the investigation but also offer more reliable results. The aerial photograph offers possibilities of detailed of the terrain and its culture suited to the need of the investigator, be he a geologist, forester, soil scientist, town planner or any other kind of specialist. The only stipulation is that the specialist must know what he is looking for and how the information, which he is seeking, appears in the aerial photograph. The result of this qualitative analysis will depend upon the specialist's level of specialization and experience. The quantitative analysis which involves measurements of linear distances, angles and height differences between terrain objects under investigation as well as the preparation of base maps will only be possible if the geometry of photographs and technique of photogrammetry are understood. These notes will deal with those aspects of photogrammetry, which should form the base of photogrammetric knowledge for photo-interpreters in the disciplines of geology, forestry, soil sciences and town planning. GEOMETRY OF AERIAL PHOTOGRAPHS Projection In order to understand the geometric qualities of a photograph it is necessary to understand what projection means in terms of geometry. In the examples given (Fig. 1a, 1b &1c) the triangle ABC and the line LL' on which the projection is made are in the same plane. a) Parallel Projection In this projection, the projecting rays are parallel. The triangle ABC is projected on the LL'. The projection of the triangle is `abc'. The projection rays Aa, Bb, Cc, are all parallel in this case. (Fig.1a)

33

Fig. 1(a) Fig. 1(b) Fig. 1 (c) b) Orthogonal Projection Fig. 1b gives an example; in this case the projecting rays are all perpendicular to the line LL'. This is a special case of parallel projection Maps are an orthogonal projection of the ground on a certain scale. The advantage of this projection is that the distances, angles and areas in the plane are independent of the elevation differences of the objects. c) Central Projection Fig. 1c shows a central projection. The projecting rays Aa, Bb, Cc, pass through one point O, called the Projection Centre or Perspective Centre. The image projected by a lens system is treated as a central projection, (though stringy it is not, as the lens is not a single point). TILT It is the angle between the optical axis of the camera and the plumb line. It is also the angle between the ground plane and the photo plane. Tilt can be resolved into two components, one in the direction of flight (the X-axis) and the other perpendicular to it (the Y-axis). I. The component about the Y-axis, i.e. in the direction of X is called Longitudinal Tilt

or X-tilt or Fore and Aft Tilt or Tip. It is denoted by letter ϕ (Phi). II. The component about the X-axis, i.e. in the direction of Y is called Lateral Tilt or Y-

Tilt or simple Tilt. It is donated by letter ω (Omega)

34

Fig. 2

In Fig. 2 the vertical ON through the perspective center meets the photo plane at point `n' called the Photo nadir point and the ground plane at point N called the Ground nadir point. These points are also called Plumb Point. The foot of the perpendicular (p) from O on the photo plane is called Principle Point. The length of this perpendicular (op) is called Principle Distance. The approximate position of the principal point of a photograph is determined by joining the opposite fiducial marks (or collimating marks) (Fig.3a). Line joining opposite fiducial marks is known as fiducial axis. The point of intersection of the fiducial axes is called fiducial centre (f) and is, for practical purposes, coincident with the principal point (p) in a well-adjusted camera.

Fig. 3(a) Fig.3 (b)

35

Figure 4. Locations of the principal point (PP), the nadir (n) and the isocenter (i) on a tilted vertical aerial photograph.

Figure 5. Diagram of a tilted photograph illustrating the location of the principal point (PP), the nadir (n), the isocenter (i), the axis of tilt, and the direction of tilt (up and down sides). Reasons for Photo Tilt:

36

I. Atmospheric conditions (air pockets or currents) II. Human error of the pilot fails to maintain a steady flight, III. Imperfections in the camera mounting, etc. SWING Swing is the angle measured in the plane of the photograph between the fiducial axis in the direction of flight and the actual flight line (Fig. 3b). The angle is denoted by (Kappa). ELEMENTARY MATHEMATICAL CONCEPTS An aerial photograph, as already discussed, is a central perspective. In an ideal case of an absolutely vertical photograph of a completely flat terrain the aerial photograph will be geometrically the same as the corresponding map of the area. However, because of tilt of the photograph and relief variation of ground photographed, an aerial photograph differs geometrically from the map of the corresponding area. The central perspective (a case of central projection) is characterized by the fact that all straight lines joining corresponding points, i.e., straight lines joining object points to their corresponding images, pass through one point. This point is known as the perspective centre. Fig. 6 illustrates this relationship.

Fig. 6

37

Straight lines Aa', Bb' etc., joining corresponding points, e.g. A on the ground and a' its image in the image plane (negative plane), are known as perspective rays and pass through the perspective centre . A plane between the perspective centre and the object is known as a positive plane. The consideration of a positive plane does not involve any significant geometrical change in the relationship. The purpose of Photogrammetry is to produce an orthogonal projection of the image of the object from its central projection(i.e., from perspective pictures-photographs) by using the geometrical links between the object and its photo-image at the moment of exposure. Properties of Central Projection To study the properties of an aerial photograph it is necessary to understand the geometry of central (perspective projection). Some properties of this projection will be dealt with in what follows : Plane I can be considered as ground plane and Plane II as positive plane (photograph), see Fig. 7. a) AB, the line of intersection of the object and image planes is known as the axis of

homology. It is also known as the axis of perspective. b) SN and Sp are the perpendiculars from S on to the planes I and II intersecting plane I

at N and P and plane II at n and p respectively. The bisector of the angle PSN meets the principle line (see subpara (f) EF in plane I and EH in plane II in I and i. These points are ISOCENTRE is that angle measured at I in plane I are the same as corresponding angles measured at i in plane II.

It is very important to remember that angles are true at the isocentres only when the ground is flat.

c) A plane parallel to plane I and passing through the perspective centre `S' cuts the plane

II in a line CD that is known as the horizon line. Horizon Line and Axis of homology are always parallel to each other. All horizontal lines parallel to the horizon line or axis of homology are termed as plate parallel. The plate parallel passing through the isocentre `i' is termed the isometric parallel or isoline. It can be proved that this is the only parallel along which the scale = f/H, i.e., the same as in the case of a vertical photograph.

A line through the perspective centre perpendicular to the principal plane (see para (f) is called axis of tilt.

d) Image of all objects infinitely distant on the right of AB will be formed on the horizon

line. Points on the horizon line are known as vanishing points.

38

Images of parallel lines in the ground plane converge to a vanishing point. Thus, images of all lines parallel to the principle line in the object plane will converge to a vanishing point H, the point of intersection of horizon line and the principal line in the photo-plane.

It is a fundamental property of perspective projection that a line in one plane projects as a line in the other plane, the two lines meeting at the axis of homology, e.g. line XY in ground plane projects as X'Y' in photo plane.

e) A plane, which is perpendicular to both the planes and passes through the perspective

centre is known as the principal plane. The lines of intersection of this plane with the two planes are known as Principal lines. EF and EH are principal lines. Fig. 5a represent the principal plane. The angle `0' between the principal lines is the angle between the perspective planes. When this angle `0' is equal to Zero the plane II can be considered as a vertical photograph. In normal vertical photography this angle should not exceed 3 or 4 grades.

f) Any point in plane I such as X has a corresponding position X' in plane II. Such pairs are

called homologous points. g) If two planes are projectively related as in Fig. 8 certain important relationships exist

between the corresponding details in plane I and II. Lines O'1, O'2, O'3 and O'4, on plane II are the images of lines O1, O2, O3 & O4 in plane I. Let there be another line UV in plane I which cuts the lines O1, O2, etc. in 1', 2', 3' and 4' then it can be shown that 12/13 1'2'/1'3' -------- = ---------- = r 24/34 2'4'/3'4' This ratio is known as the Anharmonic ratio (or Cross-ratio) of the four distances. Because of the constancy of this ratio, a unique position can be found for the line UV in plane II as well so that 1' falls on line O'1, 2' falls on line O'2 and so on. This property is used in graphical rectification.

39

Fig. 7 Geometry of a tilted photograph

fig: 8

40

SCALE, GROUND COVERAGE AND RESOLUTION OF AERIAL PHOTOS,

TILT AND RELIEF DISPLACEMENT SCALE OF PHOTOGRAPHS Scale is the relationship between distance on a map or photo and the actual ground distance. Scale is represented in two says: a) Equating different units of measurement on map and ground, i.e. 1 inch = 1 Mile, 64

inches = 1 Mile b) As R.F. (representative fraction) in which the numerator is unity, e.g. 1:10,000 or

1/10,000, which means 1 unit on the map or photo, represents 10,000 units on the ground.

Methods of scale determination In decreasing order of accuracy these are : i) By establishing the relation of photo to ground: If the distance between the same two points on the photo as well as on the ground can be measured, R.F. can be set up : Photo distance R.F. = ---------------- Ground distance ii) By establishing the relation of photo to ground with the help of a map : If the distance between two points on a photo which can be located on the map as well, is measured, the horizontal measurements of these distance form a ratio, which when multiplied by the R.F. of the map gives the R.F. of the photo. If `g' were the ground distance between two points, `m' the map distance and `p' the photo distance then R.F. of map is m/g and R.F. of photo scale is p/g. R.F. of photo p/g p _____________ = --- = - R.F. of map m/g m p

41

Hence R.F. of photo = --------x R.F. of map m iii) By establishing the relation between focal length of the Camera and the flying altitude : In a true vertical photograph of flat terrain the scale of photograph is the ratio f/H. In figure 9(a) distance `AB' is imaged as `ab' on the photo. photo distance Scale of the photo = ----------------- ground distance ab = -- AB f = - .... From similar H triangles OAB & Oab

Fig. 9(a)

If the terrain is not flat, the scale of the photograph is not uniform. In Fig. 9(b), Hm is the flying height above the average height of the terrain photographed. Then the average scale of the photograph = f/Hm.

42

Fig. 9 (b) The scale of photo for a point A which is at a height of `h' metres/ft. above the average ground level f = -------- (the units of focal length and the heights Hm-h being in the same terms). Similarly, the scale for another point B which is at a vertical distance `d' metres/ft. below the average terrain level f = ----- Hm+d Thus the scale of photograph is not uniform if there is irregular terrain. We can determine either the average scale of the photograph as a whole or the scale of the photograph at a particular point or elevation. Higher areas will be on a larger scale than that of lower valleys. In tilted photograph the scale is not constant. It is constant along any plate parallel (if the ground is flat). The scale along isometric parallel (discussed earlier) is true, i.e., equal to f/H. The scale increases continuously on the nadir point side of the isocentre and decreases continuously towards the principal point side of it.

43

Thus we arrive at an important result: The scale of aerial photograph changes irregularly due to height difference in the terrain but continuously due to inclination of the camera axis. Resolution Resolution of aerial photograph is expressed in lines pair per millimeter, i.e., nos. of lines and equal size gap can be resolved. We can get approximately 20 lines pair per mm on the scale of the original negative. For example if the original scale of negative is 1:10,000, then the ground resolution will be - 1 1 --------X10, 000 mm = ----- x 10,000 mm = 25 cm. on the ground. (20+ 20) 40 IMAGE DISPLACEMENT On a planimetric map all features/details are shown in their correct horizontal position on a certain scale. This is not so in the case of aerial photographs due to image, displacement or distortion. A disturbance of the principle of geometry is called displacement/distortion. There are three major sources of displacement/distortion, which are due to : Optical or photographic deficiencies, i.e. lens distortion and aberration; relief variation of the object photographed and tilt of the camera axis at the moment of exposure. (a) Lens distortion Fig. 10 shows distortion due to a lens. Object point O is imaged at I' instead of its correct position I on the image plane. d is the image displacement in this case. In the modern aerial camera lens this type of distortion is negligible.

Fig. 10: Lens distortion

44

(b) Image displacement due to relief Relief is the most significant source of image displacement. In Fig. 11, O is the camera station. NA' is a flat plain on which stands a tower AB with its base at B. The image of B on the truly vertical positive photographic plane is b. This is the correct planimetric position (orthogonal) of the image of the tower AB. Top A is imaged at `a'. The image of A is thus displaced from its correct planimetric position b, as `A' is vertically above `B', on the photograph. This shift of `a' from `b' represented by distance ba is called relief displacement. Let h be the height of the tower, H the flying height above the datum plane, n and N is the photo and ground nadir points.

Fig. 11 na The scale of photo along na is ---- NA' ab The scale of photo along ab is ---- BA' Since the photograph is truly vertical and datum plane is truly horizontal the scale will be constant. ab na Hence ------ = ------- BA' NA'

45

ab BA' or ---- = --------- ........(1) na NA' From similar triangles ONA' and ABA' BA' h ---- = ------, NA' H ab h Equation (1) becomes ---- = --- na H h Hence ab = na.------- H If we denote `ab', the displacement by r’ and na, the distance between the nadir point and the image of top of the object by r then we can write the above equation as h r’ = r. ------- H From this relation we conclude : I. Relief displacement increases with increasing value of `r' i.e., it is zero at plumb

point and maximum at the edges of the photograph. II. Smaller the height of the object, smaller is the displacement and vice versa. If h =

o, i.e. for objects in the datum plane there is no displacement. III. With increasing value of `H' i.e. with highflying heights the displacement decreases.

The satellite pictures can thus be considered having very low relief displacement. It can also be proved that the relief displacement is radial from the plumb point. While relief displacement constitutes a source of error in the measurement of horizontal distances on aerial photographs, it is the characteristics that make it possible to study overlapping photographs stereoscopically and in the determination of height differences between objects photographed. (c) Image displacement due to tilt

46

I. Flat Terrain - Let O (Fig. 12) be the perspective center I and II be the positive

planes for a truly vertical and tilted photographs respectively. The figure shows a cross-section in the principal plane. For a point `A' which appears at a' in I and at a in II, the displacement is equal to ia' - ia. It can be shown that it is equal to

ia2 . Sin Ø ----------------------- f - ia.Sin Ø and is radial from the isocentre. For a point b in plane II (Fig. 13) which does not lie in the principal plane and is that an angle with principal line at the isocentre `i', the tilt displacement which is still radial from the isocentre can be shown to be equal to

ib2 . Sin Ø.Cos2 Ø ib' - ib = ----------------------------- f - ib.Sin Ø. Cos Ø

Fig 12 Fig. 13 The displacement due to tilt is outward from the isocentre when the point is on the nadir point side of the isometric parallel and inward when on the principal point side. If the tilt is small n and i will be closer to p and, therefore, for near vertical photographs we assume that the relief displacement is radial from the principal point. (Principal point coinciding with the nadir point and the isocentre for all practical purposes) and the displacement due to tilt is negligible. This assumption is valid for all graphical methods of plotting, mean height of relief being less than 10% of the flying height.

47

The only mark easily available on the photograph, is the principal point, which can be easily plotted and is convenient to use. The isocentre or the plumb point, though easy to define is difficult to locate on the photograph. II. Accidented Terrain - We know that displacement due to relief is radial from the

plumb point and that displacement due to tilt, in case of flat terrain, is radial from the isocentre. There is, however, no such point on the photograph where angles are true to the corresponding angles on the ground in the case of accidented terrain, i.e. terrain in which there are elevational differences.

48

STEREO VISION, STEREOMODEL, AND STEREOSCOPES For deriving maximum benefit from photographs they are normally studies stereoscopically. A pair of photographs taken from two camera stations but covering some common area constitutes a stereoscopic pair which when viewed in a certain manner gives an impression as if a three dimensional model of the common area is being seen (Fig.14). The basis of this subjective impression is dealt in the end of this lesson. Depth Perception Human beings can distinguish depth instinctively. However, there are many aids to depth perception, for instance, closer objects partly cover distant objects or distant objects appear smaller than similar objects nearby. These aids apply to monocular vision. For short distances binocular vision is more important and is of interest to Photogrammetrists, for it is binocular vision which enables us to obtain a spatial impression of a MODEL formed by two photographs of an object (or objects) taken from different view points. Normally, our eyes give us two slightly different views, which are fused physiologically by the brain, and result in a sensation of seeing a model having three dimensions. This three-dimensional effect, due to binocular vision, is very limited however, decreasing rapidly beyond a viewing distance of one metre. Thus it may be concluded that binocular vision is primarily an aid in controlling and directing the movements of one's limbs. A small percentage of the people do not have the facility of binocular vision and no amount of training will give it to them. Unfortunately, there is no known physical aid to provide stereoscopic sight to such person who does not possess it naturally, but training can help those having weak fusion. Requirements of Stereoscopic Photographs If, instead of looking at the original scene, we observe photos of that scene taken from two different view points, we can, under suitable conditions, obtain a three dimensional impression from the two dimensional photos. This impression may be very similar to the impression given by the original scene, but in practice this is rarely so. In order to produce a spatial model, the two photographs of a scene must fulfill certain conditions: a) The camera (spatial) axes should be approximately in one plane, though the eyes can

accommodate the difference to a limited degree. b) The ratio B/H, in which B is the distance between the exposure stations and H is the

distance between an object point and the line joining the two stations, must have an appropriate value. In aerial photogrammetry this ratio is called the base-height ratio.

49

If this ratio is too small say smaller than 0.02, we can obtain a fusion of the two pictures, but the depth impression will not be stronger than if only one photograph was used. The ideal value of B/H is not known, but is probably not far from 0.25. In photogrammetry, values upto 2 are used, although depending on the object, sometimes much greater values may be appropriated.

c) The scale of the two photographs should be approximately the same. Difference upto

15% may, however, be successfully accommodated. For continuous observation and measurements, differences greater than 5% may be disadvantageous.

d) Each photograph of the pair should be viewed by one eye only, i.e., each eye should have

a different view of the common overlay area. The brightness of both the photographs should be similar. Such a pair of photograph is known as stereoscopic pair or stereogram. Stereoscopic vertical photography is the most commonly used one in aerial survey. The terrain is covered with strips of photographs. Overlap between two photographs in the same strip varies from 55 to 90%. Overlap of adjacent strips varies from 5 to 55%. The most usual overlaps are, in the strip, 60% and between two adjacent strips, 25%. Binocular Observation of Stereoscopic Photographs Accommodation and convergence If we have a pair of stereoscopic photographs in front of us, on paper, glass plates or projected with projectors and they are oriented in such a way that epipolar lines are situated in the way described before we can observe them in different ways. In order to evaluate the different ways of observation, we have to use the terms accommodation and convergence. Accommodation refers to focussing of eye-lens to see objects sharply at different distances. An un-accommodated eye is considered to be focussed at infinity. Convergence refers to the directing of lines of sight (i.e., the optical axes) of the two eyes to the same point. The optical axis of the eye can be changed in direction by rotating the eye in its socket. The angle the eye base subtends at the point is called angle of convergence or parallactic angle. Normal reading distance is 250 mm, i.e., while reading we accommodate and converge the eyes at this distance. As the eye-base is one an average about 65 mm (2.5 inches) for human eye, the angle of convergence them is approximately 16 degrees. (The line joining the nodes of the eyes is called eye-base or the interocular or interpupillary distance (Fig. 15).

50

The relation between the accommodation distance (d) and angle of convergence (in radians) is given by E = --- d E being the interpupillary distance

Fig. 14 Fig. 15 Normally accommodation and convergence are automatically linked up. If we look at a point at a certain distance, accommodation and convergence are set for that distance. We can disconnect this link but not without much strain on eyes. A lot of practice is required for accommodation at a distance other than the distance of convergence. There are three ways of observation of stereoscopic photographs: a) Observation with Crossed eye axes

This involves looking with the right eye at the left photograph and with the left eye at the right photograph (Fig. 16(a). The convergence and accommodation are at two different distances, and this type of observation is, therefore, very tiring. Large photographs can be used conveniently by this method, but due to strain on the eye, this method is not used in practice.

b) Observation with parallel eye axes

This method is possible without any optical aids, but is tiring as well as the eyes are converged on infinity, yet accommodating at approximately 250 mm (Fig. 16(b). It is less tiresome if positive lenses are placed between the eyes and the photographs so that the photos are placed at the focal length of the lenses. The accommodation

51

then corresponds with the convergence and the eyes are viewing naturally. The `pocket-stereoscope' was developed on this principle.

c) Observation with convergent eye-axes

When the accommodation and convergence are at the same distance the viewing is least tiring and this is the normal method of viewing. But in order to view the photos stereoscopically they must be superimposed, such that the point A and the corresponding point A' on the other photo lie at the point of convergence (Fig.16(c).

The images have to be separated so that left eye sees only the left hand photographs and the right eye only the right hand photograph. The resulting stereoscopic perception is similar to that of normal 3 dimensional perception. The separation may be achieved by colour filters or by polarized filters. There is an interesting phenomenon in Stereoscopy. In viewing terrain in aerial photography a reversal of the relief is sometimes obtained by the eyes. Such a phenomenon is known as pseudoscopic illusion or Pseudoscopy. Such an impression can be obtained by viewing the photos with crossed eye axes. Sometimes, viewing with the shadows in case of excessive relief (e.g. hills) away from the observer can also result in pseudoscopy. So, in the initial stages, to avoid pseudoscopic view, it is desirable to view the photographs with shadows of objects falling towards the observer. Separation by colour filters The photos are either projected or printed in two different colours. By placing a filter of the same colour over each eye corresponding picture is observed by one eye only. In practice this problem is difficult to solve completely. The human eye is sensitive for light with wavelength from 400 to 720 millimicrons (mu). Fig. 17 shows the spectral sensitivity curve of eye. The vertex lies at about 560 mu. A possibility for separation of the two superimposed images would be to use filters of which one cuts off all wave length over 560 mu (its colour would be blue-green) and the other all under 560 mu (orange-red). The image projected in orange-red can be observed with an orange-red glass in front of the eye. With the blue-green image it is just opposite. This means, we have on one of our retina at bluish image from one projector, and on the other a red one from the other projector. We seem to be able to fuse these different images to one stereoscopic white-black image. In the case of anaglyphs printed on paper the condition is different from that described above. The two images are printed in red and blue. The eye covered with the red filter sees both the red image is indistinguishable and only the blue image is visible as varying shades of gray. Similarly the eye covered with the blue filter seems the red image

52

only. If the spectacles are reversed, we see the LH photograph with the right eye and vice versa. A pseudoscopic image will result. Separation by polarized filters Light has the characteristics of a wave motion in which the waves vibrate in all possible planes perpendicular to the direction of prorogation. These are called transverse waves. It is possible to analyze the transverse waves into separate components along two axes perpendicular to each other and to the direction of propagation by means of filters. For stereoscopic vision the filters are placed so that polarized light rays forming the left image are at right angles to the light rays forming the right image. There are several advantages in using polarized light : - light loss is about 50% only in both projections, - there is not colour contrast between the two picture, and - it is possible to use colour photography on this principle. However, there is one big disadvantage in using the method, which has so far prevented its use in photogrammetry. With the type of plotting instrument, which uses this system, it is important, that the screen on which the image is projected be diffuse, so that it can be viewed equally well from all directions but a diffuse surface acts as a depolarizer and so no stereoscopic image would be apparent. STEREOSCOPES The function of a stereoscope is to deflect normally converging lines-of-sight so that each eye views a different photographic image. Stereoscopes are grouped into 2 basic types : i) Lens stereoscopes ii) Mirror prism stereoscopes Pocket Stereoscope By far the most popular is the lens stereoscope commonly known as pocket stereoscope. The pocket stereoscope usually has plane-convex lens, upper side flat with a focal length of 100 mm. The rays entering the eyes are now parallel and converge at infinity and have been accommodated (focussed) at 100 mm distance (Fig. 18). Since the normal viewing distance is 250 mm, a closer view., i.e. at 100 mm result in a magnification. The magnification is then 250/100 = 2.5. More expensive types have a changeable eye base. Such a refinement is not necessary for operators with an average eye-base range of 60 to 68 mm. The pocket stereoscope is cheap, transportable, and has a large field of view. It has two big disadvantages:

53

Fig. 16 Fig. 18

Fig. 17 a) Limited magnification. Pocket stereoscopes with more than three times magnification

cannot be equipped with simple plane-convex lenses, due to the too large an increase in lens aberrations. In addition the distance between the head and the photos becomes to small for adequate illumination without undue complications.

b) The distance between corresponding points on the photos must be equal to or smaller

than the eye base. With normal size photographs this becomes difficult or impossible without bending or folding the photos.

It should not be forgotten, however, that due to the simple optical system the image quality of the pocket stereoscope is very good.

54

Mirror Stereoscope The two above mentioned drawbacks have led to the development of the mirror stereoscope. The normal size photos (23 cm x 23 cm) can be separated and seen under the stereoscope without folding them. The path of the bundle of rays has been diverted and brought to the eyes at 65 mm separation. This is achieved by reflecting mirrors. Normally the distance between corresponding points is kept at 240 mm so that photographs are placed separately, i.e., it effectively increases the eye base from 65 mm to 240 mm. As in pocket stereoscope the picture must be at the focal plane of the lenses in order to have convergence at infinity. The mirrors M1 are placed in such a way that the picture distance via the small mirrors M2 (generally prisms) become equal to the focal length of the lens, usually 300 mm (Fig. 19(a). This gives approximately 250/300 = 0.8 x magnification, or rather reduction the picture observed. to magnify the image additional oculars of magnification 3x to 8x can be used over the prisms or a lens placed before each prism (See Fig. 19(b) giving a magnification of about 1.8x.

Fig. 19 (a) Fig. 19(b) SUBJECTIVE SPATIAL MODEL The subjective spatial model observed with a stereoscope when photographs having overlap are viewed is termed as Stereo-model. If one observes the ground from an aeroplane one does not see a spatial model. The eye base is so small (65 mm average) compared with the flying height of the aeroplane that the two-retine image is virtually the same. Therefore, there is no true comparison between the natural view and the stereoscopic view of a model. It may be assumed that we see natural relief if we observe an object with a normal base-height ratio. In the light of what has been said before, i.e. that binocular vision was mainly an aid in controlling the movements of the limbs, we could say that a normal base-height ratio is near about 65/250, i.e. about 1/4 to 1:1 or even 1:0.6, this could lead to the conclusion that the stereoscopic image formed by aerial photographs is always different and distorted. However, there are other factors, which influence the subjective model.

55

Assume that the photographs are taken with a vertical optical axis and that they are observed flat on a table, oriented according to the epipolar rays. a) The first difference is that the eye-base has been changed from say 800 mm to 65 mm.

This change only alters the scale of the model and the two views remain similar in every other respect.

b) The second difference is that the photographs are observed at a distance, which is not

equal to the principal distance. This, not only, alters the magnification of the model but simultaneously alters the ratio between the x, y scales against the z scale. We get an affine flattened model if this distance is smaller than the principal distance and exaggerated if it is greater than the principal distance. This corresponds with what one finds in practice.

c) The third difference is that our eyes are moved away from the vertical through the

principal points. This produces deformations difficult to construct or visualize in a diagram.

d) The fourth difference is that one of the photographs is moved during observation, so

that the corresponding points are seem vertically. This shift is equal to the stereoscopic parallax (P), and makes the rays from the corresponding points to the respective observation. However, this parallelism renders the construction of the spatial image impossible as it means that the spatial model should be formed at infinity. In practice the image does not appear at infinity but an indeterminate distance varying from 250 mm to 1 meter according to the personal idiosyncrasies of the operator.

e) Lastly the shape of the object, the shadows, the natural association of the observed

data and relative distance all influence the process of depth perception.

56

MEASUREMENT OF HEIGHT FROM AERIAL PHOTOS, PARALLAX AND PARALLAX MEASUREMENT

Parallax The feasibility of finding height differences of objects with the help of measurement on photographs of the area concerned is the most important quality of photographs. This is achieved by measuring parallaxes on the photographs. What then is parallax ? The term parallax is applied to the apparent change in the position of an object caused by change of position in the observer. The term is widely used in optics, astronomy and other sciences and has different significance in each case. In photogrammetry we are generally concerned with stereoscopic parallax. The aerial camera does not take aerial photographs continuously but takes them at certain exposure intervals. suppose instead of the negative film there was a ground glass on which ground images could be seen, then it will be seen on changing with respect to the camera frame. Consider that at one instant the airplane is at 01, vertically above a point P. The image of P will appear at p on the ground-glass (Fig.20). After sometime when the plane is at 02, it will appear at p'. This shift pp' in the position of the image of P on the ground glass is the parallax of P. Similarly for any other point Q will be qq'.

Fig. 20

57

Again it will be seen that images of the higher points in the terrain will move across the ground glass more rapidly than the images of lower points in the valley. Thus the separation (Parallax) of the images of a higher point would be more than the separation (parallax) of lower points (during the same interval of time). That means, points at higher elevation exhibit a greater parallax than those at the lower elevation. X- and Y- Parallaxes In Fig. 21, P1, p2' and p2 p1' are the photo bases of the left hand and right hand photographs respectively. a1 and a2 are the corresponding images of an object point A p1 a1 and p2 a2 can be resolved into two mutually perpendicular directions - one along the direction of flight (X-direction) and the other perpendicular to it (Y-direction). Then, if X1, Y1 and X2, Y2 are the resolved parts of p1 a1 and p2 a2 respectively in the two directions.

Fig. 21: Parallax of principal points X1 - (-X2) = (X1 + X2) is the X- parallax or absolute stereoscopic parallax or horizontal parallax and is defined as the algebraic difference in the direction of the air base of the distances of the two images of an object from their respective principal points. (Note:- Minus sign is given to X2 as the distance from the principal point is measure in the negative X-direction, i.e., opposite the flight direction). Similarly (Y1 - Y2) is called Y-Parallax or Vertical Parallax. If the paired photographs are assumed to be vertical and taken from equal altitude above the datum, the Y-Parallax is absent.

58

Parallax of principal points If we transfer the principal point P1 of the left hand photograph of a stereo pair on the right hand photograph at p1' (Fig. 21), then by definition p2 p1' is the parallax of principal point of the left photograph and is the distance between the exposure stations (air-base) on the scale of the right hand photograph. Similarly, p1 p2' is the parallax of the principal point of the right hand photo and represents the air-base on the scale of the left hand photograph. If the terrain is flat and flight is level and flight altitude does not change the scale of the two photos will be exactly same and hence the two photo-bases will be exactly equal. It is, therefore, simple to find out the parallax of either of the principal points by measurements on the photographs, provided the assumption do not deviate much from ideal situation, viz. i) focal length in both cases same, which is always the case ii) flying height is the same iii) optical axis vertical In practice we tolerate tilts of about 3 degrees; of course, flying heights are within reasonable limits. Parallax Difference Assuming that there is no tilt and flight is level, two photographs are taken with image of an object point A a1 and a2 on them respectively. If the two photographs are put on top of the other with their principal point’s p1 and p2 and flight direction in coincidence, then by definition a2 a1 is the absolute stereoscopic parallax (Fig. 22a).

Fig. 22(a) Fig. 22(b) If now we put the pair of photographs under a stereoscope for fusion, they will have to be separated at a convenient distance p1p2 say a distance represented by `S' (Fig. 22b).

59

The parallax of A, PA = p1 p2 - a1 a2 = S -a1 a2 Similarly parallax of another point Q, PQ = S - q1 q2 Considering `A' as the reference point, the parallax difference between `A' and `Q' is ∆p = PQ - PA = a1 a2 - q1 q2 In practice, direct measurement of parallax is seldom done, instead we measure the parallax difference ( ∆ p) with the help of parallax bar or parallax wedge. Generally, graduations on parallax bar are marked in such a way that if the separation between corresponding images decreases (i.e. `d' decreasing - Fig. 22b), the reading on parallax bar increases - the point with larger parallax gives a higher reading, and correspond to a point of higher elevation. In such a case the parallax difference ∆p = (Parallax bar reading for Q - Parallax bar reading for A) = q1 q2 - a1 a2 Parallax Formula Starting with the assumption that: I. photographs are free from tilt, II. the flight altitude above the datum remains unchanged, III. the photographs are a central projections, with centre of projection at the

perspective centre, i.e., there is no lens distortion, IV. there is no distortion in the photographic material

60

Fig. 23

We have from figure 23: O1 and O2 is the air-base B at a vertical distance of ZA and ZQ above terrain points A and Q respectively; a1 and a2 are the corresponding images of about points A on the photograph. Focal length of the aerial camera lens is f. From O2 draw a line O2 a1' parallel to O1 a1. Then, by definition parallax of A is : PA = a2 a1' From similar triangles O2 a2 a1' and A O2 O1 ZA B B ---- = ----- = --- f a2a1' PA B.f ZA = _____ ........(1) PA B.f Similarly for a point Q, ZQ = ------- ........ (2) PQ From equations (1) and (2) 1 1 ZA - ZQ = B.f ( ----- - ----- ) PA PQ

61

PQ - PA = B.f --------------- PA.PQ B.f PQ - PA = ---- . X ---------- PA PQ PQ - PA = ZA . --------- PQ .... by substituting for B.f from equation(1) ----- PA ∆p = ZA . ----------- .............................(3) PA + ∆p where p = PQ - PA Let hA and hQ be the mean sea-level heights of object points A and Q respectively. Then ZA + hA = Flying height of the aircraft above datum plane (MSL) = ZQ + hQ ZA - ZQ = hQ - hA = differences of heights between terrain points Q and A = h. Equation (3) now can be written as ZA. ∆ p h = -------- ......................(4) PA + p Relation (4) is the fundamental parallax equation. Equation (4) can be put in the form (by cross-multiplying and rearranging) PA . h

62

∆p = ----------- ........................(5) ZA - h We have assumed that these parallax equations (equations (4) and (5) are valid only when the photography is vertical and the flight is level. However, it may be applied for small variations from these ideal conditions. If h values are small (e.g. height of tree, embankment), the simplified formulae ZA . ∆p h = ---------- PA can be used and similarly PA . h ∆ p = --------- ZA For reasons of convenience for the absolute parallax of the principal point of the left hand photograph (PA) the length of photo-base on right hand photograph, is commonly measured and substituted in the solution of parallax equations. For near vertical photographs or relatively flat terrain, the use of average photo-base of the stereo-pair gives reasonably accurate results. For formulae then become Z . ∆p bm . h h = --------- and ∆p = ------------ bm + p Z - h and the approximate formulae can be written as Z bm h = ---. p and ∆p = ---- . h bm Z Where Z is the average flying height above the terrain and bm the average photo-base. Image displacement due to tilt of any one of the two photographs causes false parallax across the overlap. Similarly the slope of the air base affects the parallaxes. As such the parallaxes observed on such photographs are burdened with errors. For the two points, between which the height difference is required, if not far from each other, the effect on parallax due to tilt and inclination of air base is nearly the same and will cancel each other.

63

Floating Marks Floating marks are also known as the Measuring marks as these are used for precise measurement on stereo photographs. These are defined as pairs of identical reference marks when viewed stereoscopically in conjunction with a photographic overlap combine to form a single floating image. If we put two dots, A1, A2 (Fig. 24) about a mm in diameter at a distance of about 65 mm on a piece of paper and see them under a pocket stereoscope (the eye base kept parallel to the line joining the dots), they will fuse into one dot. Now if we put another set of dots B1, B2 close to them and such that the line joining them is parallel to the eye base and spaced closer than the first set and seen under stereoscope, we find that this set also fuses into one mark, but floating i.e. higher above the first one. The vertical distance `AB' is known as stereoscopic depth. That means if the mark has a different parallax from that of its surroundings, it will appear higher or lower. But if no parallax exists between the dots and the object images, the fused dot appears in contact with the fused image. At this moment quite accurate measurement may be made of the distance between these reference marks. Measurement of Parallax Difference

Parallax difference can be measured with ordinary ruler but it cannot give

accurate/precise results. For accurate results the principle of floating marks is used in parallax bar or parallax wedge Thus the function of these stereo meters is to measure changes in parallax that are too small to be determined with the ordinary rulers. Parallax bar

A parallax bar consists of two glass plates, A and B, engraved with identical measuring marks, connected by a bar (Fig. 25). The separation S between the marks can be changed by a micrometer screw. M, graduated so as to give reading upto 0.01 mm. Glass A can be shifted along the rod and can be clamped by screw C. Graduation on the bar are arbitrary and do not refer to the actual separation S of the measuring marks. The graduations on the micrometer and the bar are usually numbered increasingly as the distance between corresponding points, i.e., the separation S, is decreasing.

Fig. 24 Fig. 25

64

Parallax wedge This is a sheet of transparent material with no converging rows of dots. The wedge

is slid backwards and forwards in `Y' direction until two dots fuse as one dot on the ground, the reading then being noted. The dots are numbered in accordance with corresponding parallax values. (Fig. 26).

Fig. 26

65

STEREO IMAGES FROM SATELLITES Stereo images can be obtained from satellites also in the same way as the are collected from an aircraft, and the study is called as satellite photogrammetry or satellitegrammetry. In case of stereo photography from satellites the basic mathematical and geometrical aspects remain same as in aerial photogrammetry. However in case of satellite photogrammetry, the flying height is of order of some hundreds of kilometers in comparison with, a few kilometers, in aerial photogrammetry. As a result of this, the earth's curvature plays more than usual role in geometry. The ratio of earth's relief with flying height also becomes too small in satellite photogrammetry, thus seriously effecting the perception of relief in stereo images, obtained from satellites. Cameras as well as scanners have been used in satellites to obtain stereo images. During early stages of manned and unmanned satellite missions, small 70mm cameras were used to take photographs, at very small scale and low resolution. Later, attempts were made to obtain better stereo images from satellites using large format cameras and scanners, which are described, in the following sections. STEREO IMAGES FROM LARGE FORMAT CAMERA The Large Format Camera (LFC), a special purpose camera with high resolution and high geometric fidelity, was flown in NASA Space Shuttle mission in October 1984. The LFC in Space shuttle mission is a precision cartographic camera, with image format 230 X 460 mm and focal length 305 mm and advanced image motion compensation mechanism to account for the shuttle ground velocity of 27000 km per hour. The photographs were taken from varying flying height of 235 to 375 km above earth. Stereoscopic coverage was taken with variable overlap ( 20 to 80 percent) in the same way as in aerial photography. The photographs retained sharp details even after enlargement and were found to have high cartographic value. The original photos were at scale of 0.8 million to 1.2 million. On another mission, a 230 X 230 mm format photogrammetric camera with 305 mm focal length lens was also used with a colour infrared film. The ground resolution was found to be of order of 25 meters. Skylab, a manned satellite, carried an Earth Terrain Camera (ETC, S-190B Experiment) had a 457 mm focal length lens and acquired photographs with a ground coverage of 109 by 109 km. The photographs were collected in normal color, black and white and IR colour at a scale of 1:950,000on 11.4 cm square film. The overlap was 60 percent, which is well suited for photogrammetric purposes. The satellite images taken from Russian Satellite during mid eighties, from a camera with focal length of one meter and ground resolution of about two meters are now available for the most part of world. The photographs taken from satellites at altitude of few hundred kilometers on a photographic film are seriously effected by atmospheric haze. Therefore either color infrared, or panchromatic films are used with minus blue filter to cut the blue haze.

66

ACROSS TRACK SCANNERS FOR STEREO IMAGING The common optical mechanical scanners, using a mirror, sweeping the small area (pixel) being viewed, across the flight path are known as Across Track Scanners (and also as whiskbroom scanners). The Multi Spectral Scanner (MSS) used in Landsat series of satellite comes under category. The orbit of Landsat and other similar satellites is such that, at equator adjoining images have about 14 percent overlap. This overlap increases towards Polar Regions to as much as 85 percent. In such areas of high overlaps, the two images taken from adjoining satellite path could be used as stereo images. The base to height ratio varies from 0.174 at the equator to 0.031 at polar regions. In such images obtained by across track scanners, the relief displacement is in cross track scan direction, (outwards from the center of the scan line in each scan line). However, due to limited ground resolution, and unsuitable base to height ratio, the relief perception is not suitable for cartographic application. ALONG TRACK SCANNERS WITH STEERING CAPABLITY The along track scanners are also known as pushbroom scanners. They consist of linear arrays of numerous Charge Coupled Device (CCD). High Resolution Visible (HRV) imaging system used in SPOT satellite series and PAN (a panchromatic camera) in IRS-1C satellite belong to these type. The satellites are provided with steering capability to tilt the sensor with an angle towards right or left. Thus, off nadir view could be obtained, covering the same area from a different orbit. The amount of tilt is variable to plan for a suitable look angle from available orbits to cover the area of interest. In these scanners also, the relief displacement is normal to the ground track. The stereo images are found to be suitable for medium scale mapping. Many researchers have found that SPOT image (PAN) with 10 meter resolution in stereoscopic mode, can be used for topographic mapping at 1:50,000 scale with 20 meter contour interval. Geometry of PAN camera in IRS-1C Satellite This uses reflective optics along with 4096-element CCD linear array (7 micron X 7 micron) for imaging. A special arrangement, comprising of an isosceles prism reflector, is used for covering full swath of 70 km. spatial resolution at nadir is 5.8m. The swath steering range is ñ 26 o, the step size of ñ 0.09 and the repeatability for stereo coverage is 0.1o. The data is collected in panchromatic mode having spectral band 0.50 µm to 0.75 µm. The space segment specifications are given in Table 1. PAN data is encoded in 6-bits.

67

Table-1. Space segment specifications of IRS-1C Orbit type Polar Sun synchronous

Altitude 817 km

Inclination 98.69o

Distance between adjacent traces 117.5 km

Repetivity for LISS-3 24 days

Repetivity for WIFS 5 days

Off-nadir coverage ñ 26o for PAN 398 km.

Stereo viewing capability 5 days Geometry of HRV camera in SPOT Satellite series SPOT pushbroom scanning system does not employ a scanning mirror, rather, it employs a linear array of CCDs arranged side by side along a line perpendicular to the satellite orbit track. The HRV contains four CCD subarrays of 6000 elements each, acquiring data in panchromatic mode to record 10m resolution data. Three 3000 element subarrays are employed in the multispectral mode at 20 m resolution. HRV optical system has a plane mirror, which can be rotated to either side by, ground command, through an angle of ñ 27o (in 45 steps of 0.6o each). This allows each instrument to image any point within a strip extending 475 km to either side of the satellite ground track. At latitude of 45o, there are six possible occasions during the 26-day orbit cycle on which successive day stereo coverage may b e obtained. At the equator, only two stereo viewing opportunities on successive days are possible. The base height ratio also varies with latitude from approximately 0.50 or 45o to approximately 0.75 at the equator. If stereoscopic coverage need not be acquired on successive day basis, the range of possible viewing geometry greatly increases. The specifications of the orbit of the satellite is given in Table 2. Table-2. Orbital characteristics of SPOT series satellites

Altitude 832 km. Orbital period (min) 101 Inclination (degrees) 98.7 Equatorial crossing time 10.30 AM (local sun time) Sensors HRV Repetivity 26 days Stereo viewing capability 5 days

68

ALONG TRACK SCANNER WITH FORE & AFT LOOKING Some pushbroom scanners employed fore and aft looking mode with the help of external mirror to produce stereo images. This external mirror attachments provides forward looking and an aft looking channel in addition to the normal downward looking (nadir looking) channels. In Multispectral Electro-optical Imaging Scanner (MEIS II), which was developed for the Canada Centre for Remote Sensing, acquired data in eight spectral bands ranging from 0.39 µm to µm1.1 µm, with an IFOV of 0.7 mrad and a total field of view of 40o. This was the first airborne pushbroom scanner to be used operationally. RADAR STEREO IMAGES RADAR is acronym to Radio Detection And Ranging where microwave pulses are transmitted to illuminate the terrain and back-scattered pulses from the terrain are received to produce images of the terrain. The operating principle and other details are discussed separately in other sections. Stereo images can be acquired using Radar also in the same way as the across track scanners are used i.e. images are collected from two adjacent paths. However, because the Radar side lighting effect will be reversed on the two images of the stereo pair, stereoscopic viewing is little difficult using this technique. Radar can acquire stereo images by varying the flying height also, if it is airborne, or by having two antennae with two different look angles. There are so many other parameters are to be considered in reproducing stereo models from these type of Radar stereo pairs other than from photogrmammetric point of view. Measuring parallax and feature heights and the related study, from these images is called Radargrammetry. Among Space borne Radar systems, SIR-B and SIR-C, the space shuttle imaging Radars and RADRASAT, a Canadian satellite presently in the orbit, collected stereo images. SIR-B Radar experimental mission flown during October 1984, collected stereo images in L-band, HH polarization, with varying look angles from 150 to 60o. The azimuth resolution was 25 m and the range resolution varied from 14 m at a look angle of 60o to 46 m at a look angle of 150. SIR-C collected images in mutli-frequency, multi-polarization modes with varying look angles ranging from 15o to 55o. Radarsat SAR is a C-band system operating with HH polarization. The system can be operated in a variety of beam selection modes providing various swath widths, resolutions and look angles. The Modular Optoelectronic Multispectral Scanner (MOMS) was developed in Germany and works like pushbroom scanner with linear array of CCD. The size of actual ground swath covered varies with pointing angle employed. At the 27o maximum value, the swath width for each instrument is 80 km. When the two instruments are pointed so as to cover adjacent image fields at nadir, the total swath width is 117 km and the two fields overlap by 3 km. While each HRV instrument is capable of collecting panchromatic and multispectral data simultaneously, resulting in four data streams, only two data streams can be transmitted at one time. thus either panchromatic or multispectral data can be transmitted over a 117 km wide swath, but not both simultaneously.

69

TABLE 4 Salient features of Future SAR systems planned for launch after (1993) Mission Year Freq. Pol. Look angle Swath Resolution (deg) (km) (m) SIR-C 1994 L, C & HH&VV 15-55 15-90 30 (USA/ X-Bands HV&VH GERMANY) VV ERS-2 1994 C-Band VV 23.5 80 25 (EUROPE) ALMAZ-2 1996 -- -- -- -- (RUSSIA) RADARSAT 1995 C-Band HH 20-40 100 25 (CANADA) 20-40 150 35 37-49 45 10 49-59 300/500 100 49-59 75 30 EOS 1998 L, C & VV,HH 15-55 30-120 30 (USA) X-Bands HV&VH 700(Scan) 15 HH&VV ENVISAT 1998 C-Band VV,HH 20-50 100-400 30 (EUROPE)

70

PRINCIPLES OF STEREOPHOTOGRAMMETRY

1. Introduction A photograph is formed by a bundle of rays through the camera lens. In stereophotogrammetry the photograph itself is not used, but the bundle of rays that made the photograph. We reconstruct such a bundle with the help of a photograph and a perspective centre. In a stereo-pair we have two such bundles covering a common area. If we reverse the process by using replicas of the lens of the taking camera and place them at the air stations O1 and O2 (Fig.1) and pictures placed in the positions with correct orientation as they existed at the time of photography and illuminated from behind by a powerful light source, we can recreate the terrain model. This model will be formed at a distance as it physically existed at the time of photography, i.e., at a distance of flying height and on scale 1:1. In practice, it is not feasible to recreate such a model. By reducing the air base, we can reduce the model to a convenient size. The reduction in size of the model will be in the same ratio as the reduction of the air base. This principle is made use of in recreating ground model on suitable scale with the help of stereo-pairs of photographs of the terrain and a restitution instrument. 2. Inner Orientation Inner orientation is the process by which the image-forming bundle of rays for each photo is reconstructed, true in its geometry. If we put the developed negative/diapositive in a projector and place a light source behind it, we reconstruct the same bundle of rays, provided we use the same lens and the same principal distance as used for taking the photography. If we change the principal distance in the reconstruction, the geometry of the bundle is completely disturbed - the height and plan scales will differ. If the position of the principal point relative to the optical axis of the projection lens is changed, the geometry of the bundle is again disturbed. Figure 2 shows the disturbance in the geometry of the imaging rays. From the above it is clear that in order to reconstruct the true bundle of rays, the picture must be correctly positioned in the projector.

71

3. Exterior Orientation By inner orientation, the central projection in the image space is completely fixed. Its relation to the object space is not yet known. This is provided by what is known as exterior orientation. To understand this consider following :

The motions, which can be given to, a projector are: - a) 3 translations along the X, Y and Z-axes. Small shifts of the projector in these

directions are represented by bx, by and bz respectively. b) 3 rotations around these axes. The rotation around the X-axis is called transversal tilt

and is denoted by ω ; the rotation around the Y-axis is called Longitudinal tilt and is denoted by φ and the rotation around the Z-axis is called Swing and is denoted by χ.

Fig. 3 shows the six motions of a projector, which are generally known as the elements of orientation.

Fig. 1

72

Fig:2

Fig. 3: Motions of Projector

If we know the inner and outer orientation of a single photograph, the pencil of rays for all points in space are directed towards the proper object points but the object cannot be recreated. Double image photogrammetry is able to give complete information about the objects. For this we use a pair of stereoscopic bundles, of which exterior orientation is determined by 12 elements, 6 of each bundle. Exterior orientation is achieved in two steps a) Relative Orientation b) Absolute Orientation

73

Relative Orientation

It is the process of establishing the angular relationship between the two consecutive photographs as it existed at the instant of exposure. In other words, one bundle is placed relative to the other in space in the same way as they existed at the time of photography. Relative orientation is obtained when all the corresponding rays from both the projectors intersect simultaneously. The condition of intersection of such pairs of rays is attained when the X- and Y- parallaxes are zero, in a three dimensional model. In Fig. 4 corresponding rays O1 p1 and O2 P2 meet the projection plane at P1 and P2. The separation of the images, P1P2, is the parallax, which can be resolved into two components: (i) In the flight direction (X-axis) and is called X-parallax (∆x) (ii) In a direction perpendicular to the flight direction (Y-axis) and is called Y-parallax ( ∆ Y) Y-parallax is eliminated if the rays O1P1 and O2P2 are made to lie in the epipolar plane of point P. This can be achieved by operating any of the elements of the projector by, bz, ϖ ,φ or χ (except bx) depending upon the location of point P in the model. The X-parallax can be eliminated either by lowering or raising the projection plane or by reducing or enlarging the instrument base (which represent the air base) O1O2 (Fig.5). X-parallax is, thus concerned with height or scale only. From Projective Geometry it can be proved that if 5 pairs of corresponding rays intersect simultaneously in the overlapping are, all the corresponding rays of the two bundles will intersect. Therefore, it involves eliminating Y-parallax at five points. From practical point of view the solution or removal of y-parallax at the chosen points in the model must be brought about by those elements of the orientation, which have maximum influence on y-parallax. Model Points: From the six patterns of the effects of the linear and rotational movements of the orientation elements, in the projected images it is possible to select suitable locations for the elimination of y-parallaxes in the model. Model points 1 to 6 (Fig. 6) are found most suitable for this purpose. Five of these are usually chosen for relative orientation and the sixth point for a check. According to general practice we consider two nadir points (points 1 and 2) and four corner points symmetrically located (points 3,4,5 and 6) as model points.

74

By changing the base length by `bx' element of the projectors we change the scale of the model only. Therefore, element `bx' cannot be used for relative orientation. We are now left with 10 elements (5 of each projector) to affect the relative orientation. Out of these, we have to select five, which are most suited for elimination of y-parallax.

Fig. 4 Fig. 5 Fig.6: Model Points Although many combinations are possible, the two most commonly used ones are :(figures 1 and 2 with an element of orientation indicate the particular projector used) I. χ1, χ2, φ 1,φ2, and ω1 (or ω2), i.e. only rotational elements of both the projectors. II. All the five elements of one projector, the other projector remaining fixed, e.g. χ, φ,

ω, bz and by of one projector or the other. Absolute Orientation:

Absolute Orientation of a model involves horizontalisation and scaling which so far have remained undetermined. (i) Levelling the model:

At lease 3 points within the overlap area suitably located (e.g. points 3, 4 and 6 in Fig. 6 or other suitably located) must be known in height (above m.s.l.). Both the cameras are suitably moved simultaneously until height differences in the model correspond to true height differences. A fourth point of known height (e.g. point No. 5) provides a check on leveling.

(ii) Scaling the model : At least two points of known planimetric co-ordinates, well separated in the

overlapping area must be known. By comparing the true distance between these points and that measured in the model, a scale factor can be calculated and the instrument base corrected to set the model at the desired scale.

It is now clear that for absolute orientation at least the complete coordinates of 2 points (i.e. X, Y and Z) and the height (i.e. Z coordinate) of a third point must be known.

75

The necessity of orientation is inherent in all photogrammetric survey work where approximate solutions do not suffice,, i.e., where high precision is required.

MODEL DEFORMATION

When corresponding rays from two projectors do not intersect correctly, there are residual positional errors in X,Y and Z coordinates of points after absolute orientation of the model framed. These results in a model, which is deformed. The errors in X and Y directions are small and have negligible effect on planimetry. The errors in X-parallaxes of points show up as appreciable height differences (the heights are measured in absolute terms instead of being plotted). The parallax formula derived in earlier, which is used for determination of heights, is derived on the assumptions that: a) the photographs are central projections. In actual reality, it is not so strictly. The lens

system introduces distortions. b) the base `B' is exactly horizontal. In actual practice the two exposure stations are

seldom at same altitude. c) c) the camera axis is vertical at the time of exposure. It is seldom the case in

practice; at best the axis is near-vertical. and thus X-parallaxes results from - I. non-fulfillment of the above stated conditions, II. the paper on which photograph is made is not being dimensionally stable, and III. The two photographs forming a model are seldom in their exact relative positions

with respect to each other and this shortcoming introduces additional parallaxes, which vary in different parts of the model and give rise to various deformations.

Thus a model of a flat ground under these circumstances would appear warped. Deformation due to Orientation Elements Consider a perfectly oriented model, in which corresponding rays 01 P and 02 P (Fig.7 intersect at P on the projection plane. One of the orientation elements of the right hand projector is disturbed slightly, introducing X-parallax. It will be seen that the projected image of P will shift to, say, P' on the projection plane, while that from the left-hand projector remains at P. The corresponding rays instead of intersecting at P now intersect at P' which is different level in the model. Thus, the X-parallax, X, introduced

76

by slight movement of one projector results in the deformation Z at P. The pattern and extent of deformation can best be studied by keeping one projector (say, Left-hand projector) fixed and by considering the effect of one element of exterior orientation of the other projector at a time. a) Influence of bx:

Fig. 8 shows the effect of bx movement of the right hand projector to right hand side. There is a constant shift of all the points and results in the datum plane being lowered down.

b) Influence of by:

Fig.9 shows the effect of by. It gives a constant shift in all points in Y-direction and has no effect on X- parallax, and hence does not result in any deformation.

c) Influence of bz:

This results in reduction of enlargement of the scale of the projection. In Fig. 10 the X and Y components of the shifts are shown. The model shows tilted in X-direction.

d) Influence of k : It will be seen that there is no X-parallax at points 1 and 2 (Fig.11), while that at

points 3 and 4 is equal and opposite to that at points 5 and 6. The model will, therefore, be deformed with one end elevated and the other depressed about the axis joining 1 and 2.

e) Influence of ω:

The X-parallax at points 3 and 5 is equal in magnitude and opposite in direction. Consequently one point will come up and the other lowered. There is no X-parallax in points 1,2,4 and 6. The deformation is, therefore, a twist and has the shape of a hyperbolic-paraboloid (Fig. 12)

f) Influence of φ:

The effect of φ is shown in Fig. 13. The X- parallaxes at points 1,3,5 are equal and at points 2,4,6 are also equal but in the section 1-3-5 these are more than in section 2-4-6. The variation of X- parallax is quadratic (and not linear) in X- direction. The deformation has the shape of a parabolic cylinder .

77

Model deformation is particularly important for observations with a parallax bar. In this case the photographs are both placed flat on the table. The only orientation carried out is, by the elements k and 'by. The remaining three elements, and 'bz, which are not used in the orientation procedure cause model deformation. As such the model that we view under the stereoscope is deformed and the heights obtained by parallax bar may be burdened with errors. From the deformation patterns explained in the above figures, it will be seen that the deformation is linear in Y-direction in all cases. This property can be made use of in constructing a correction graph with the help of a few height control points in the model area for adjusting the discrepancies in heights obtained by parallax bar.

Fig.7 Fig. 11: Influence of χ

Fig. 8: bx alters the datum height Fig.12: Influence of ω (Model is not deformed)

78

Fig.9: by has no effect on model form Fig.13: Influence of φ

Fig.10: Influence of bz

79

SATELLITE PHOTOGRAMMETRY

Satellite photogrammetry imaging-systems designed for photogrammetric geodesy can be characterized as stereoscopic imaging systems capable of producing data from which a photogrammetrical can determine 3-dimensional coordinates, and related topographic information. The users’ requirements must play a key role in the design of an earth, moon or planetary mapping system. Before attempting to specify a system, which would collect data that would permit useful topographic mapping from space of the earth or extraterrestrial bodies, it is necessary to examine some of the essential factors involved. 1. A topographic map contains three kinds of information (Doyle 1973).

2. Content – the cultural and natural features represented on the map

3. Horizontal location – the reference graticule, grid, datum.

4. Elevation – spot heights, contour lines, profiles and elevations.

A satellite photogrammetry imaging-system designed for mapping must be able to

provide all three kinds of data. Map content is determined, in part, by photographic resolution and

scale, or more directly from resolved distance on the ground (resolution). It is difficult to establish a linear relation between map scale and resolution required, because some features, like roads and railroads, must be shown on a map regardless of its scale. Hence, these features are nearly independent of map scale; and therefore resolution required is not linearly related to scale. On the moon and planets, however, cultural features have, so far, not been detected so that a linear relation seems more justified. Considering the photograph as a base (photo mosaic or orthophotograph) for the map, a useful criterion can be produced.

The smallest feature, which can be depicted on a map, is assumed to

have a least dimension of 0.25-mm (.010 inch). In order for an object to be photographically identifiable it must be imaged by about 5 resolution

80

elements. It follows; then, that the resolution required for photography can be estimated as:

Ground Resolution = 0.2 x 0.25 mm x map scale number or Rg = 5 x 10-5Sm where; Rg = ground resolution required (metres) for the sensor Sm = map scale number The second kind of map information is horizontal location in an

absolute coordinate system. For mapping the earth, this is generally provided by reference to ground control points. But such control does not exist on the other planets. However, orbital-tracking data can provide spacecraft position to a high order of accuracy. When this is coupled with precise data on attitude, and time of exposure, an independent means of determining absolute horizontal location is available. This removes the need for ground control. Within a stereoscopic model, or a photogrammetric triangulation, the internal (relative) positional accuracy is approximately equal to the product of the scale of the photograph and the error in measuring, or identifying a point on the picture:

σp = Sp σm

Where; σp = standard error in relative position of a ground point σm = precision of measuring an image point Sp = photo-scale number An imaging system designed for photogrammetric surveys must be concerned with two basic categories. The first is the metric characteristics of the system and spacecraft, which are as follows.

81

1. Coordinate (X, Y, Z) on the orbit as a function of time. If the Global Positioning System is used to determine the orbit, an accuracy of ± 10 m may be expected.

2. Time (tt) of exposure correlated with the coordinates. One millisecond

precision (.001 sec) was achieved with Apollo 15, 16 and 17. 3. Attitude – Orientation angles (ω, φ, K) can be obtained from stellar

sensors such as those on Apollo 15, 16, and 17 with accuracy of the order of 5” to 10”.

4. Altimetric distance – Radar and laser distance measuring devices that are

aimed down the optical axis of the mapping camera produce distance from the spacecraft to the ground with a precision below 1 m.

5. Calibration – Determination of the radial and decentering distortion

should reduce distortion errors to less than 10 micrometers in the image plane of the sensor.

The second category is the geometrical configuration of the sensor. This takes into account the stereoscopic coverage of the surface, considering both forward and side-lap of the imagery. Also, the mode of operation whether it be vertical or convergent and its overall effect on precision in extracting relative height measurements (σh) from the stereoscopic pairs is important. ELEMENTS OF SATELLITE PHOTOGRAMMETRY In satellite photogrammetry, the perspective centres of the imaging systems may be assumed to be on the satellite’s orbit. It follows, then, that the satellite must be observed to determine the imaging system’s position in the orbit as a function of time. Examples of equipment used for observing satellites are the tracking stations of Deep Space Network and the Space-flight Tracking and Data Network. These stations are distributed around the globe.

82

These tracking stations determine direction, distance (range) and/or radial velocity (range rate) of the object being tracked. These data are converted, by a process called orbit determination, to a set of six constants, called “orbital elements”, that specify the orbit. The orbital elements most commonly used are so-called “Eulerian elements”. The satellite’s coordinates are then usually the radius vector r, whose components are the radial distances from geocenter, and the true anomaly f. For satellite photogrammetry, the Eulerian elements are usually replaced by another set: the location and velocity of the satellite at a specific time. The satellite’s coordinates are then its rectangular Cartesian coordinates. Thus the coordinates (X, Y, Z) of the perspective centres of the camera (imaging system) along an orbital and are not independent quantities, but instead are functions of the six orbital elements and time. A list of coordinates and corresponding times is called an Ephemeris. The other important quantities in satellite photogrammetry are the three angles of orientation (attitude) of the camera at the time tI of exposure. In summary, the tracking data are converted to an ephemeris of (X, Y, Z) coordinates as a function of time t. The steller camera provides three orientation angles based on the star’s positions. These are the six elements required to solve a photogrammetric problem. They may be used as first approximations in strip and block adjustments.

83

BASICS OF RADARGRAMMETRY

BASIC GEOMETRIC CONCEPTS

Radargrammetry initially requires a mathematical model relating an object point in some three-dimensional world coordinate system to the time and range coordinates, t and r, measurable from the side-looking radar record. Radargrammetry begins with the SLR image, not with the raw electronic signals acquired by the antenna. A second radargrammetric issue is the so-called sensor model that relates the radar recording to time t and slant range r. Each specific SLR system will have a unique model. DEFINITIONS

Let us assume first that we deal with a radar image as presented on

film. The coordinates along the SLR film are denoted by x, the y coordinates are defined across the film. The origin of the x coordinate may be defined arbitrarily; for example, at the beginning of the flight line. The coordinates relate to the time at which an object point has been imaged. The x coordinates axis is called the along-track or azimuthal axis. The y-axis is the cross-tract or range direction. The x coordinates are proportional to the time of flight, the y coordinate relates to range between the antenna and an object or target. VIEWING GEOMETRY

Viewing geometry is a team used to describe the basic imaging geometry of a sensor without consideration of imaging errors due to platform motion, system defects, the atmosphere, and so on. Thus, only an idealized SLR image is considered. Figure.1 reminds us of the manner in which an electromagnetic pulse illuminates the terrain and creates an image line. It is evident that ground objects will be imaged into locations on the image line as a function of their distance from the antenna. This permits us to define the radar “projection” and relate it to the geometry of a map or camera photograph, as in Figure 2.

84

For the present purpose, a map at scale 1:1 can be assumed to be an orthogonal projection of the object or terrain. This is defined by the fact that the lines connecting image and object are all straight, parallel, and normal to the projection plane (see Figure 2 a). Similarly, conventional camera photography presents a central perspective projection; that is, the projection lines connecting image and object points are all straight and all pass through one point, namely, the projection center (see Figure 2 b). The SLR projection (at scale 1:1) is basically different. The projection lines in this case are concentric circles around the antenna (see Figure 2 c). The basic property of the SLR projection, which sometimes also is referred to as range projection, leads to a number of differences between SLR imagery and an orthogonal projection of the imaged object. IMAGE OF A DISTANCE Consider first the image o9f a distance Eg, extending from point A to point B in the object space (Figure 3). The distance Eg would be represented in its correct dimension; but in a slant range projection, we would obtain Es. Having given a slant range presentation, the scale number f was defined as a constant for the imagery. If, however, the scale number of a slant range presentation is defined as the ratio between a distance in the image and the corresponding distance on the ground, then we obtain from the distance eg and es at the image scale (es = Es/f, ef = Eg/f). IMAGE OF A VERTICAL STRUCTURE – RELIEF DISPLACEMENT

Let us now consider a point, A, that does not lie in the datum or

reference plane above which the flying height, H, is measured (Figure 4) but is situated on top of a vertical structure of some height. It is obvious that, in both the ground range and slant range presentations, relief displacement, pg and ps is introduced for any point not in the reference or datum plane. From figure 3.8 we can conclude that the auxiliary angle α is nearly equal to the depression angle θ. Therefore, we obtain, in ground range presentation, and with slant ranges.

85

IMAGE OF A VERTICAL STRUCTURE – LAY OVER Another interesting fact to note is the direction of the relief

displacement. From aerial photography, interpreters are used to the fact that vertical structures, like houses, fall away from the nadir point. In radar the situation is the opposite: relief displacement is towards the nadir line, N. If AB in Figure 5 is a solid vertical structure, then A, B will be its image in a ground range presentation. This is inverted when compared to photography: the top of the structure is closer to the nadir than the bottom, and we obtain the so-called radar lay over. IMAGE OF A VERTICAL STRUCTURE – SHADOWS

Finally, there are shadows in SLR images. From Figure 4 we see that

no reflections will be received at the antenna during the time that signals should be returning from the area between points A and S but no objects are hit between these two points. Because the radiation is blocked by the vertical structure, no reflections are received from the ground points between A and S. On the image, area AS will be black, called a shadow of the vertical structure. The length of this shadow area is given sg in the ground range and as ss in the slant range presentations. We find Sg = h/tan θ And Ss = h/tan θ From these formulas we easily can see that shadows get longer if the structure is at smaller depression angles (if the aircraft flies at a lower altitude, H, or the distance to the object is larger) or if the height, h, increases. IMAGE OF A SLOPE

Let us now consider the images of slopes. Because the relief

displacement is directed towards the nadir line, slopes directed towards the antenna will appear “foreshortened”, “laid over”, or, as a limiting case between foreshortening and lay over, a slope might appear as a line in the image. This last possibility exists only for slopes that are parallel to the flight line and occurs if the slope angle equals the elevation angle = 900 - θ.

86

Three possibilities exist for slopes facing away from the flight line. First, they may be in the radar shadow; second, the radiation may only just strike the slope: or finally, it may be fully imaged. Slopes facing away from the flight line always will show a longer image than in an orthogonal map projection. SQUINT

Radar pulses do not propagate along a plane vertical to the antenna’s

longitudinal axis but along a conical surface of which the antenna is the axis. This always will occur if the real aperture antenna is end-fed, or if a synthetic aperture is created using nonzero Doppler frequencies. Radar engineers often denote any imaging mode by “squint”, in which imaging is systematically not done in a plane normal to the flight direction. We therefore may find the concept of squinted-mode imaging used for imaging with an antenna that is swung around a vertical axis. This can be accomplished only with real apertures; with synthetic aperture the “beam” always is a cone with the velocity vector is as its axis. The cone degenerates to a plane when the zero Doppler frequency is used for image formation.

87

PLOTTING INSTRUMENTS - SIMPLE AND STEREO PLOTTERS Introduction

In this chapter we will describe few simple photogrammetric instruments, designed for plotting detail from photographs in preparing base maps for natural resources survey. We will not go in detail but cover the basic principles and characteristics of these instruments. These simple instruments are classified into two broad categories : a) Instruments for plotting planimetry b) Instruments for plotting planimetry and altimetry Simple Plotting Instruments for Planimetry

These instruments are essentially tracing devices by means of which planimetric detail may be compiled on control points or new details inserted on existing maps. They incorporate means of changing scale and in some types provision is made for an approximate rectification. SKETCHMASTERS Principle of Sketchmaster:

The simplest way to get a rectified image of a photograph with tilt is by sketchmaster. The instrument is sometimes called single photo plotter. The rectification is done optically. Sketchmaster consists of an optical device, which allows one eye to receive two superimposed image, one from the photograph and other from the manuscript. Fig. 1 shows the principle of sketchmaster. In looking downwards through the aperture the image of the photograph appears to be in a plane directly below the eye through a semi-transparent mirror M, at a distance equal to the distance of the photo from the eye along the optical path EMP. If we place a map at this distance, the image of the photograph will coincide with the map in respect to common points, provided the scale of the two are equal. The effect of tilt can be corrected by tilting or rotating either the photograph or the map plane or both, relative to each other.

88

Fig. 1: Principle of sketchmaster

If the photo image and map do not lie in the same plane, optical parallax will result. Scale error and parallax can be eliminated by inserting a lens, positive or negative, depending upon the required enlargement or reduction, between the map and the mirror. The function of this lens is to bring the map image into a sharp focus at the plane of the photo-image. Zeiss Aero-Sketchmaster: The main parts of the instrument are: i) Plate carrier which can be tilted or rotated in any direction ii) ii) A double prism with semi-silvered surface, which permits viewing of, photograph

and map sheet simultaneously through an oblique eye-piece (Fig. 2). Together with the prism the plate carrier is vertically adjustable above the map sheet on the tabletop, i.e. photo-prism and map-prism distances can be varied to suit the scale solution.

89

Fig. 2: Aero - Sketchmaster

By tilting the photograph and changing the prism-photo and prism-map distance relationships, the operator sees the photo-image superimposed on the map. To compensate for the difference between the eye-photo and eye-map distances, the instrument is inserted in the double prism in between the prism-map and prism-photo. The operator sees coincidence between the plotted control points on the map and corresponding photo points. For flat areas this number can be larger than 3, but for hilly terrain it is generally not possible to obtain coincidence on more than 3 points unless all points are at the same elevation. So, in practice the instrument is moved from one group of 3 points to another and details are traced directly on map sheet. The instrument is also provided with smoked glasses with light transmitting power of 25%, 50% and 75%, which can be inserted in the double prism near the lenses. These facilitate balancing of illumination of the photo and map. Better effect is perceived, if they are equally illuminated. Use of Sketchmasters:

Sketchmasters are used for transferring details from vertical or near vertical

photographs to map by tracing, in areas where many changes have taken place since the map last prepared. They are also used for original compilation or small-scale planimetric charts provided the relief of the terrain is low (depending upon the mapping scale accuracy desired).

90

SIMPLE PLOTTING INSTRUMENTS FOR PLANIMETRY AND ALTIMETRY A. Simple Stereometer Type of Instruments: These instruments provide capability for making stereoscopic measurements on aerial photographs. From parallax formula ZA + ∆P h = -------------------- PA + ∆P It is clear that height difference is the function of the parallax for a known point, parallax difference and the flying altitude above the known point. In instruments, which provide plotting of the altimetry, the device for measuring parallax difference is essential. Such instruments are essentially a combination of (i) Stereoscope and (ii) Stereometer. The difference in elevation can be obtained by the parallax difference. Model deformations and image displacements, due to tilt and relief, are inherent as the instruments only make copy of one of the photographs. No provision is made for their correction. This is a serious draw back. The condition for plotting is only that the parallax bar is equipped with a pencil holder. Parallel guidance mechanism is provided in some instruments to facilitate movement of the parallax bar parallel to itself (Fig. 3).

Fig. 3: Principle of Stereometer - type of Instruments In some instruments y-displacement mechanism on one of the floating marks is also provided so that the bar can be used on parallel guidance mechanism and allowance for removal of y-parallax without moving the photos. The pencil point moves on the map sheet. The plotting is done by following the detail with floating mark and detail is traced by pencil. Contours are plotted by moving the

91

parallax bar in such a way that the floating point remains in contact with the terrain (without changing the spacing of the floating marks). STEREOPRET (Zeiss)

This instrument is based on the above principal and is the most complete outfit of

this type. The pencil point is fixed to the left hand floating mark. Thus, whatever is the height of the point in question, the distance between the pencil point and the left hand floating mark remains same. That means for planimetry, a copy of the left hand photo is made. The photographs are fixed under glass plates on a parallel-guided double plate carrier. Parallax bar, stereoscope and illumination units are fixed but the plate carrier is moved to scan the overlap. Right hand plate carrier can be moved with the help of screw to eliminate Y-Parallax, if any. Pantograph is provided to facilitate plotting at scales other than photo scale. For contours and heights the right floating mark is used. Stereopret has no special advantage over other instruments except that it is easier to operate. Accuracies of map plotted on these instruments

The maps prepared in these instruments are burdened with the following errors :

i) Relief displacement ii) Tilt displacement iii) Distortion due to lens, and iv) Distortion in paper prints Errors due to relief displacement always are present on such maps. Errors due to tilt can be avoided by using rectified prints. Lenses used in modern cameras exhibit negligible distortion but old Eagle IX type camera lenses show larger distortion which influence the accuracy in plan as well as in height. Shrinkage of paper prints is overcome by using prints on stable material, e.g. D/W paper, correctostat paper, the size of which does not alter with temperature and humidity. The best is to use diapositives on glass or stable film. They also allow a better definition of the images. Tilt, shrinkage of paper print and lens distortion also influence the x-parallax and hence the heights. Effects of tilt and difference in scale of the stereo-photos result in deformation of the model. Heights obtained will be reliable only if rectified prints on stable material is used. The operation of these simple stereometer-type of instruments involves computations and the results are burdened with errors. Though, these instruments are

92

simple and inexpensive, it is not justified to use these in practice. More complicated instruments, though expensive, are much easier to operate and give better precision and output in shorter time. PLOTTING INSTRUMENTS - PRECISION STEREOPLOTER

For the purpose of measurement, mapping and aerial triangulation work, various

types of photogrammetric Stereoplotters are available in wide range of variety in design, application, precision and cost. The most basic types are - i) Optical Projection type:

In this the photograph diapositive is optically projector through a

photogrammetrically matching lens. In older models (e.g. Multiplex of William Ross of U.K., and Balplex of Baushc and Lomb of USA, both being obsolete now), the original photograph of 23 cm x 23 cm was reduced 4 to 3 times for easy handling in a smaller size projector, resulting into loss of details and precision. In another model (Kelsh Plotter of keuffel and Easer of USA) the original size diapositive is used but only a small part of diapositive is illuminated for observation in order to reduce the weight of projector. Scale of stereo model, being large in this case, and a pantograph is used for plotting at reduced scale. The instruments of optical projection type are usually poor in precision (2nd or 3rd order rating) and require a semi dark room conditions for working. However, they are usually universal type of instruments, which can be used for all photogrammetric application e.g. aerial triangulation (by bridging, and stereo plotting. ii) Mechanical Projection type

The more recent type of steroplotters have been designed as Mechanical Projection

type, in which, function of optical projection are performed by mechanical components, e.g. a universal joint replaces the perspective centre, a straight rod replaces the patch of ray of light from image point through perspective centre, distance between universal joint and image plane represents the principal distance, and intersection of two space rods as the point of stereo model. Tilt and translatory movements are provided or simulated mechanically (e.g. Zeiss parallelogram in Wild Autograph A-7 and Zeiss Stereoplanigraph C-8, both of universal type, I order rating, and out of production now). The II order precision instruments in this category are Wild Autograph B-8, Zeiss Planitop, Planicart, Kern PG-2, Wild Aviograph, all being non universal type. The I order non universal type instrument in this category are Wild Autograph A-10, Wild Aviomap, Zeiss Planimat. iii) Analytical Projection type

The new generation of photogrammetric stereo plotters is of Analytical Projection

type, using a digital computer. The position of image points in photograph is sensed by transducers and the data like principal distance, and coordinates of ground control points are fed to the computer by operator. Computation of values of all the orientation elements

93

is carried out by computer and incorporated in the stereo model. The accuracy of all such systems is of I order as the mechanical design is very simple, and complicated computations are carried out by computer including Relative and Absolute Orientation as well as Aerial Triangulation. This category includes the instruments Wild Aviolyt AC-1 and BC-2, Zeiss Planicom C-100, Optical Mechanical Italiana (OMI) -APC-4, Kern DSR-11, DSR-15. They are all of universal type as they can carry out all photogrammetric applications. The instruments of optical or mechanical projection type or analytical type incorporate an exact solution to the problem of reconstruction of stereo model, and come under the category of precision instruments. Very high precision can be attained with photogrammetric instruments but such instruments are very expensive, i.e., cost is high, maintenance is costly, require air-conditioned rooms and specially trained staff for operation and maintenance. These precision instruments are grouped into 3 categories; a few important ones are mentioned herewith planimetric and height accuracies. 1. Analogue Instruments

a) First order instruments - Wild Autograph A-7 Zeiss Stereoplanigraph C-8 Wild Autograph A-8 The first two of these are universal instruments. Planimetric precision of these instruments is about 10-15 µm on the scale of negative and the height precision is about 0.15%. H. b) Second order instruments - Wild Aviograph B-8 Wild Aviograph B-9 Kern PG-2 Planitop The planimetric precision of this group is about 15 to 30 on the negative scale and the height precision is about 0.025%. H. 2. Analytical Instruments Planicomp C-100, P-3 (Carl Zeiss) Aviolyt BC-1, AC-1, BC-4 (Wild) DSR – 11, DSR-15 (Kern) Intermap (Intergraph)

94

3. Auto correlator Type – Digital photogrammetric workstation available from Carl Zeiss, Leica, Intergraph.

The planimetric and height precision in such instruments depends upon the resolution (pixel size) at which the photographs are scanned, as well as the precision of camera calibration and ground control points.

Present trend in photogrammetric instrumentation is toward digital photogrammetric workstations. Due to increase in cost of precision optics and mechanical systems, many large manufactures have stopped producing precision optical mechanical and even analytical systems. However, the digital photogrammetric systems require the data in digital form, for which aerial photographs are to be scanned on a high precision photogrammetric scanner (pixel size of 5 to 25 microns). The cost of such scanners is often prohibitive and even more than the cost of digital photogrammetric plotters

95

AERIAL TRIANGULATION, CONTROL AND MAPPING Introduction

To produce an accurate map, a number of carefully determined points are required to be fixed in the area to be mapped. These points form a framework on which the survey of physical details in the area is based and are termed as control. This control prevents the accumulation of any system of errors in the measuring operations. Control may be horizontal (plan) or vertical or both. Horizontal control is needed to maintain correct scale, position and orientation of map while vertical control is needed for location of contours or leveling of a stereo model in photogrammetry. A coordinate system is essential to define the position of a point. Commonly used systems are : a) Spherical coordinate system: The position of a point is denoted in terms of latitude ( λ ) and Longitude (L). b) Rectangular plane co-ordinate system: The position of a point is defined in terms of

Easting Northing from a reference point is called the origin of co-ordinates. For heights the datum is the mean sea level. In mathematical terms the coordinates of a point is generally denoted as (X, Y, Z). The first two of the three are for planimetric coordinates and the third is for height. GROUND METHODS OF PROVISION OF CONTROL

There are many methods of provision of horizontal and vertical control by

ground methods. The principles of important ones are being mentioned below : a) Horizontal or Planimetric Control

i) Triangulation It is the process of measuring the angles of a chain or network of triangles formed by stations marked on the surface of the earth. The calculation involved is the trigonometrical proposition.

Sin A Sin B Sin C ------- = ------ = ------- a b c Hence if any side of the triangle is known, the triangle can be solved. In decreasing order of the quality, precision and instruments used, triangulation is classified as:

96

a) Primary or First Order b) Secondary or Second Order c) Tertiary or Third Order ii) Traverse

In this process the lengths and directions of azimuths of a series of connecting lines, from a point of known position to the point whose positions is to be determined, are measured.

Like triangulation, traverse is also classified as Primary, Secondary and Tertiary. iii) Astronomical fixings‚ (Astro-fix)

Latitudes and Longitudes of a point are determined by astronomical observations. The method is very fast but the degree of accuracy with which the coordinates are determined is quite low and is not suitable for accurate work. However, it is used in exploratory and reconnaissance surveys.

iv) Using Doppler or GPS

Latitudes, longitudes and heights can be fixed for a point using a Doppler or Global Positioning System using a artificial Geo-satellites. GPS are getting more popularity these days.

b) Vertical Control

Leveling is the method of providing vertical control and can be subdivided into 3 main groups.

i) Spirit Leveling

It is the process of determination of elevation of points (usually called Bench-marks) with respect to each other or with respect to a common datum by means of instruments using a spirit level or a precision pendulum. This is the most accurate method of providing vertical control.

ii) Trigonometrical Leveling

In this method the elevation difference between two points is obtained by means of observed vertical angles combined with the length of the line joining the two points. This method is combined with triangulation and traverse procedures and is the commonly used method.

iii) Barometric Leveling

The method is based on variation of the atmospheric pressure with height. Heights of points can be determined with great rapidity but the degree of

97

accuracy is quite low and is unsuitable for accurate work. It is used for preliminary reconnaissance and exploratory surveys.

DENSITY REQUIREMENS OF CONTROL FOF PHOTOGRAMMETRIC MAPPING Requirement for Stereo plotting

It has been discussed earlier that in order to fit the stereo model with the ground

at certain scale, a few control points are required. For scaling at least two plan (horizontal) control points are needed but third is needed for a check. For leveling 3 points at suitable locations are required and a fourth point for a check. Thus, we require 4 height control point and 3 plan control points. If these points are plan and height control both, then we need 4 such points located at the corners of a model. A fifth point is always required to check for the deformation. The suitable location for this point is the centre of the model. A model once oriented absolutely in these points enable other detail in the model accurately surveyed in plan as well as in height. Requirement for Rectification

Four or more planimetric control points are required at the corners of the photograph for its rectification. EXTENSION OF GROUND CONTROL BY PHOTOGRAMMETRIC METHODS

If a service dealing with natural resources inventory has to work in an area where no maps exist, it is necessary for the construction of its own base map to have a framework of control. Even if the survey is carried out in an area with a denser network of secondary or tertiary triangulation, generally it will be found that pass-points will not be in a proper location on the photograph or model. In both cases provision of control has to be made. It is now clear that whether compiling a line map from a single photograph or from a stereo pair or rectifying a photograph or preparing an aerial mosaic, a sufficiently dens network of control at suitable locations is required to orient the photograph or the stereo model. Provision of such a dense control by ground methods would be very expensive and time consuming. Fortunately, methods have been developed for the determination of this control by photogrammetric methods called. AERIAL TRIANGULATION

A small number of ground control points are, however, essential in the execution of these methods.

98

We distinguish two methods: a) Radial Triangulation -

These methods are based on angular measurements on the photographs resulting in only planimetric coordinates X and Y. The control, obtained by these methods, has proved to be highly satisfactory substitute for ground control. With only a few ground control points around the periphery of the area of survey, a network of supplemental control can be established at the desired locations‚ which is suitable for many types of mapping.

This can be done by simple methods wherein no calculation or measurement is involved. Photo-interpreter often requires the extension of control for planimetric survey. These simple methods are, therefore, very suitable for him in the making of his base maps. The methods may be: i) Graphical method ii) Slotted template method, if better accuracy is required. iii) Analytical method In case a natural resource service needs 3 coordinates, X, Y and Z, of all control points, it will be better to ask for assistance of a photogrammetric service.

b) Aerial Triangulation in Space Aerial Triangulation in Space resulting in X, Y and Z coordinates of each point. This

triangulation is carried out in precision stereo plotting instruments (or in analytical plotters) wherefrom coordinates of points fixed are obtained in machine coordinate system. These are transferred into ground system with the help of ground control points. There are three types of aerial triangulation in space:- i) Bridging Method ii I.M.T. (Independent Model Triangulation) iii) Bundle Block Adjustment (Plate Coordinates) I. Bridging Method

The another type of triangulation (Spatial Aerial Triangulation) is carried

out on Precise Photogrammetric Stereo plotting machines. This involves, reconstruction of exact stereo model, on which measurement of model coordinates of control points is carried out along the three axes. Once first stereo model in a strip is absolutely oriented, and measured, the second stereo model is constructed by orienting the third projector (photograph) with respect to the second projector, (which is already in absolute orientation), and thus the second stereo model, so generated is also in Absolute

99

Orientation. After measurement of second stereo model, the third stereo model is generated in the same way and so on. This method is knows as Bridging, and is adopted on instruments, which have the provision of constructing successive stereo models with the help of projectors. This type of instruments includes the most basic ones e.g. Multiplex, having a large number of projectors (one for each photograph) as well as more recent ones e.g. WILD Autograph A7, with only two camera projectors, used for first model with first photo on left and second photo on right projector. The second model is created by second photo on right and third photo on left projector, and third model is created with third photo on left and forth photo on right projector, and so on. The first, third, fifth models etc. are observed with some changes in observation and measuring system (known as base out condition). II. IMT Method

Another technique of spatial aerial triangulation is known as Independent Model

Triangulation, in which each stereo model, after Relative Orientation on stereo plotter, is measured independently for all the control points for X, Y and Z values. These models coordinates of all of the stereo models, are later on used to connect the stereo models to form a full strip, which is operated upon for Absolute Orientation with the help of ground control points by mathematical solution. As in this method part of work is done on stereo plotter and rest by computation, it is also known as Semi Analytical Method. III. Bundle Block Adjustment (Plate Coordinates)

In case of purely Analytical Method, the precise image coordinates are

measured on each of the photograph, by using either a stereo comparator or a mono comparator, for all the points, and entire rest of operation is done by mathematical computations, finally to yield the ground coordinates of all the control points. This involves huge amount of computations, for which a computer is indispensable. ADJUSTMENT OF ERRORS

While carrying out aerial triangulation over a large area, covered by a number of

strips, and specially while using analytical or semi analytical approach, the ground control point requirement can further be reduced, if all the strips are triangulated together in one block. This process is termed as Block Triangulation, while triangulation of each strip individually is known as Strip Triangulation. While the minimum ground control points required for Absolute Orientation is just three number of points, it is a usual practice to use much more number of ground control points for this purpose. The ;surplus points are used to check and control the accuracy. Care is taken to make sure that all the ground control points are uniformly and properly distributed throughout the area. After ground coordinates of all the points are computed, the residual errors at check points (surplus ground control points) is determined, and corrections are applied to all the points by a suitable mathematical model, in order to make the combined magnitude of residual errors to be minimum. This operation is known as

100

Adjustment of errors. In most of the adopted practices, in Analytical Aerial Triangulation, this operation is carried out along with other photogrammetric computations, and entire computing operation is known as Block Adjustment. Modern Workstations

Modern analytical Photogrammetric Instruments usually contain a microcomputer

with entire measuring, recording and data storage system with computation, analysis and error adjustment programs, and in a system. This facilitates the entire operation at a single workstation and thus much faster in delivering the final output. GROUND CONTROL FOR PHOTOGRAMMETRIC TRIANGULATION From above discussions it is clear that a few ground control points are essential to carry out photogrammetric triangulation, which ties the photogrammetric survey to the ground. Actual amount of control varies with the scale and accuracy requirements. For aerial triangulation in stereo-plotting instruments, if ground control points are available in every fifth or sixth model in each strip, they enable in determination of supplemental control points in each model. For graphical methods the density of ground control should be about 20 cm apart on the plot sheet on the scale of survey, while in slotted template method about 40% of the above is sufficient. The ground control for photogrammetric triangulation may be provided : a) By carrying out control survey on the ground viz., triangulation, traversing, leveling etc. b) By making use of trigonometrical framework existing in the country. This may often

require supplementation on the ground. c) By making use of certain types of surveyed detail appearing on the existing maps,

depending on the accuracy requirements of the photogrammetric triangulation. This type of control may be a good substitute for supplementation control for rectification, mosaicing and even for base map compilation depending on the accuracy desired.

Accurate identification and proper description of ground control on the photograph is an important step. Any error in the identification or in marking on a photograph is equivalent to an error in fixing the point on the ground. The ground points should, therefore, show up beat in plan, i.e., they should be distinctive both on the ground and photograph, e.g. road intersections; rail road crossings; fence corners; isolated huts, trees, unobscured by vegetation or shadows.

101

DIGITAL PHOTOGRAMMETRY

Digital photogrammetry or Soft Copy photogrammetry is the latest development in the field of Photogrammetry. This has a lot of similarity with the already existing analytical Photogrammetry. Until last two or three years analytical and digital Photogrammetry terms were used synonymously, since the internal processes use the same mathematical models used for solving unknowns are same in both. Also in both the systems most of the end products are same. However, there exists a considerable difference in input, viewing system, and level of automation in some areas, when we compare these two branches of photogrammetry. Analytical or computational methods were existing since a long time in Photogrammetry. However, the concepts have been realized in production mode since last three decades only, in the form of analytical stereo plotters and the most latest being digital Photogrammetry workstations. Photogrammetry as science has its use predominantly in map making. Analytical Photogrammetry

Unlike the empirical determination of unknown parameters in analogue Photogrammetry here the unknown parameters of the camera are solved mathematically and subsequently object space coordinates are computed from these parameters. The mathematical models used are based on linearized collinearity / coplanarity condition equations. External input for the solution of equations consists of camera interior orientation parameters and ground coordinates of control points and internal input consists of image coordinates of instrument itself. From these data the computer calculates in real-time, model coordinates and other forms of useful output data and then displays information on the screen. They can handle any type of photography, including vertical, tilted, low oblique, convergent, high oblique, panoramic and terrestrial photos. They can also accommodate photography from any focal length camera, and in fact can simultaneously use two photos of different focal lengths to form a model. The products from these are of superior accuracy because, since they have a capability to correct any combinations of systematic errors caused by camera lens distortion, film shrinkage or expansion, atmospheric refraction and earth curvature. In every phase of its operation, it can take advantage of redundant observations and incorporate the method of least squares into the solution of the equations. Present production oriented systems are of this type.

102

Fig.1 Analytical stereo plotter configuration

MATHEMATICAL MODEL

The basic mathematical model based on which the unknown parameters are solved is the collinearity equation as given below: xa Xa - Xo ya = R Ya - Yo ....... eq (2.1.0) -f Za - Zo where R denotes the rotation matrix of the combined rotation of x,w and q respectively around z,x and y axis, this is an orthogonal matrix R = x, w, q ....... eq (2.1.1) where x,w,q denote kappa, omega and phi respectively. R can also be expressed in matrix form as follows: m11 m12 m13 The elements of this matrix is a product m21 m22 m23 of kappa, phi, omega which needs to be m31 m32 m33 determined. X,Y,Z are the ground co-ordinates of the known point Xo,Yo,Zo the ground co-ordinate of projection center which needs to be determined on further expansion we get M11(Xa-Xo)+M12(Ya-Yo)+M13(Za-Zo) x = -f -------------------------------------------------- M31(Xa-Xo)+M32(Ya-Yo)+M33(Za-Zo)

103

..... eq (2.1.2) M21(Xa-Xo)+M21(Ya-Yo)+M23(Za-Zo) y = -f --------------------------------------------------- M31(Xa-Xo)+M32(Ya-Yo)+M33(Za-Zo) The above equation is non-linear in nature. Hence it is linearized using Tayler's series and only first order derivatives are taken for forming observation equations. COORDINATE SYSTEMS :

Here we talk about the co-ordinate systems in the context of Photogrammetric

solutions. The basic three co-ordinate systems involved are : Photographic co-ordinate system.

This is the internal reference system in the photographic camera and all image

points will be defined with respect to its axes. Every photograph employed in analytical photogrammetry contains a set of discrete points around the photo perimeter, commonly known as fiducial marks. The intersection of the lines joining the opposite marks is the fiducial centre which is taken as origin ( refer to figure 2 ).

Y

P PX C

PP PY

X

PC

Fig.2 Photo co-ordinate system

104

Model co-ordinate system.

Model coordinate system refers to the spatial coordinates of points imaged in a

stereoscopic model, which usually relates its positions with respect to the camera base or to the instrument axis. Coordinates are expressed in x, y, h (refer to figure 3).

Fig.3: Model co-ordinate system

105

Object space co-ordinate system.

Object space coordinate system refers to the coordinate system used to define the

position of points in the object space, as distinguished from the image or the model. In the context of the earth as the object we may consider one of the three coordinate systems described below. a) Geodetic coordinates of latitude, longitude and height above particular ellipsoid. b) Geocentric universal system. c) Local space rectangular system. COORDINATE TRANSFORMATION

In analytical or digital photogrammetry most of the mathematical models are used

to perform coordinate transformations. The procedure for converting one coordinate system to another is known as coordinate transformation. The procedure requires that some points have their coordinates known in both the arbitrary and the final coordinate systems. Such points are called control points.

Two-dimensional conformal coordinate transformation is applied for plain surfaces. A conformal transformation is one in which the true shape is preserved after transformation. A two dimensional conformal coordinate transformation consists of three basic steps: 1) Scale change, 2) Rotation and 3) Translation / Shift. x a b x Cx ....... eq (2.3.1) y -b a y Cy Two-dimensional affine coordinate transformation

In digital Photogrammetry both 2D and 3D coordinate transformations are used, to transform intermediatery co-ordinate systems in a process to ultimately derive terrain coordinates. Following are the affine equations, which transform from the XY comparator axis to the xy photo system: x = a1 + a2 . X + a3 . Y ....... eq (2.3.2) y = b1 + b2 . X + b3 . Y

In this case scale change is not equal in both the axis, thereby the true shape is changed.

Three-dimensional conformal coordinate transformation: It involves converting from one three-dimensional system to another. In this transformation true shape is retained. This type of coordinate transformation is essential in analytical or computational photogrammetry for two basic problems : 1. To convert coordinates of points from tilted photographic coordinate system to an equivalent vertical photographic system, which is parallel to, the ground or arbitrary object space system. 2. To form continuous 3

106

dimensional strip models from independent stereo models. A Three Dimensional Transformation is represented by following equation, where X,Y,Z are ground coordinates, x,y,z are photo coordinates, Cx,Cy,Cz are coordinates of origin and a1 to a9 are elements of transformation matrix.

X a1 a2 a3 x Cx Y = a4 a5 a6 y + Cy Z a7 a8 a9 z Cz ANALYTICAL STEREO PLOTTERS

An analytical plotter is composed of four basic components 1. Image stage system driven by servomotors and controlled by plate processors. 2. Real time mathematical processing system - for maintaining a stereo model, aiding

graphic superimposition. 3. Stereo viewing optics 4. The graphic data collection/editing system

As is necessary with all analogue stereo plotters, interior, relative and absolute orientations are also required for analytical plotters prior to going into most modes of operation. The orientation and operation of all analytical plotters are quite similar. INNER ORIENTATION:

In interior orientation, a stereo pair of diapositives with x y and x' y' fiducial

coordinate systems is placed on the measuring stages. The principal distance of the diapositives and fiducial coordinates are input to the computer. Machine image coordinates x1 y1 and x2 y2 of the diapositive fiducials are then read. This phase of operation can be aided by computer-activated servomotors, which automatically drive the measuring mark to the vicinity of the fiducials. As few as two fiducials can be measured, but more recommended and upto eight should be measured if they are available to increase redundancy. From this information the computer solves a coordinate transformation, using least square if sufficient measurements were taken, to locate the principal point of the diapositives and determine the relationships of the two photo coordinate systems with respect to the instruments image coordinate measurement systems. Corrections for shrinkage or expansion are included in the transformation. A choice of coordinate transformation is available but usually the affine or projective types are used. RELATIVE ORIENTATION:

For relative orientation xy and x'y' machine image coordinates are measured at a

minimum of five points (at least six are recommended) located in the approximate positions. Again the computer will drive measuring mark to the approximate locations, whereupon the operator makes a precise pointing. Based upon these measurements the computer calculates the elements of relative orientations using the collinearity equations. The computations are

107

performed using least squares if more than five points are involved in the solution. When relative orientation is accepted, the operator notifies the computer and orientation parameters are stored for future use. ABSOLUTE ORIENTATION:

In absolute orientation, the ground coordinates of all the control points must be

first input to the computer. For absolute orientation a minimum of two horizontal and three well-distributed vertical points are required. More than minimum is recommended. However so that a least square solution can be made. When the measurements have been taken, the computer solves a three-dimensional coordinate transformation to determine the parameter that relate the model coordinate system to ground coordinate system. An analytical plotter can be oriented considerably faster than analogue plotter and usually it can be accomplished in 10 minutes or less. DIGITAL IMAGE CORRELATION IN ANALYTICAL INSTRUMENTS:

In the past there has been considerable effort in automising the analogue or

analytical systems by image correlation. For analogue it was electronic correlation where as for analytical it was digital correlation techniques. The technique of electronic correlation, using cathode ray tubes (CRTs) to act as flying spot scanners to convert hard copy photogrammetric images into electrical signals later on actual correlation or image matching operation by hard view electrical and electronic circuitry, has been used in many photogrammetric systems of mid 60's for orthophoto production. However, with the availability of inexpensive CCD aerial array cameras, low cost video memory and digital frame stores, plug in image processing boards, high speed computing elements such as transputers and RISC could make it possible for implementing real time digital correlation analytical systems, one such system was Kern DSR-11 fitted with CCD cameras at the image intake point of the optical train. It was designed based on vertical line locus principle to extract height automatically. ANALYTICAL AEROTRIANGULATION :

Areas of mapping are mostly covered by blocks of multiple aerial photographs in

overlapping mode, because of the limitation in format size of an aerial photograph, the scale of photography required for final mapping and the area of mapping which is generally quite large. Ground control points form one of the essential input in order to solve the unknown parameters and transform a stereo model to terrain coordinate system, for mapping purposes. Minimum of two plan points and three non-linear height points are required in each stereo model. In order to provide these control points it requires enormous effort and time in the field. Also some areas are inaccessible.

Photogrammetry offers a very useful indirect / non contact method for control extension in order to generate the ground coordinates of selected photo points based on few peripheral ground control points. This method is known as aero triangulation. Aero

108

triangulation is basically a rigorous computational process which uses model / photo coordinates either from analog / analytical / all digital Photogrammetry systems as basic input. Different steps of aero triangulation: 1) Point selection. 2) Point marking (by snap marking / laser marking) 3) Model / photo coordinate observation in analytical Photogrammetry systems. 4) Preprocessing. 5) Block adjustment by independent model or bundle. ACCURACY IN AEROTRIANGULATION

The accuracy of point determination by aero triangulation is dependent on various

factors. Some of the important factors are: • Scale of aerial photography and its resolution (larger the scale and better the ground

resolution, better the accuracy in terrain terms). • Control configuration (which has greater influence on accuracy). • Use of signalized control points • Overlap conditions (poor overlapping conditions degrade the accuracy). • Measuring accuracy of the system used for observation. • Mathematical model used for block adjustment (such as polynomial, independent block

and bundle adjustment). • Natural or artificial points selected as tie points.

The theoretical accuracy achievable with optimization of above parameters and self-calibration can be of an order of 15 microns of standard deviation in planimetry, 20 microns sigma height in image scale. However in practice a standard deviation of 15 to 30 microns in XY, and 20 to 40 microns in Z is achievable generally, with an economical control configuration. However, the accuracy in Z can be improved with closer bands of height control. Although various studies do indicate different results based on various combinations and constraints, the above statement on accuracy is applicable for most of the production-oriented tasks. Salient features of analytical Aero triangulation using analytical stereo plotters: 1. Unlimited focal length (it is a general advantage in Analytical Photogrammetry as a whole). 2. Better compensation of lens distortions, film shrinkage etc.

109

3. More accurate and nearly 3 times faster than analogue because of semi automization of orientation processes. 4. Image transfer functions are used for automatic drive of floating mark to the tie points common to the adjacent models. 5. Model re-observation is done faster because of facility of restoring and retrieving of the orientation parameters and automatic driving to the already measured points. 6. Base in and out function to carry out effortlessly continuous observation of a strip with minimum number of changing photographs. (Although it existed in very few universal analogue instrument the change over from base in to out involved physically shifting of the projection centre which was equally time taking) 7. Availability of adjustment software under the same host computer so that adjustment and observations are done under same platform. DEFINTION OF A DIGITAL PHOTOGRAMMETRIC IMAGE :

A digital image ("digitised image" would be a more precise description) consists of a

two dimensional matrix G with elements g (i, j). Each element is called a pixel (word from picture elements).The row index i runs from 1 to I in steps of 1, i.e., i = 1(1)I. The corresponding index for the columns is : j = 1(1)J. Since every matrix element represents as area, we speak of image elements or pixels rather than image points. The pixels could be as small as 10 microns or even less.

The pixels g (i, j) are the information carriers. The value of a pixel depends upon the

type of recording instrument and on the computer in use. The most widely used range of values at present runs from zero to 255, a range that greatly exceeds the differentiation capabilities of the human eye. The information contained in 256 different values can be stored in eight bits (28 bit combinations) and a group of eight bits is treated as one unit, a byte, in most modern computers.

For black and white pictures the pixel values represent the gray values or densities (usually with black as zero and white as 255). For colour pictures we have three image matrices with the same ranges, i.e., we speak of an image block with three layers. If a digital image is to be used for Photogrammetric purposes we require a relation between pixel position and a xy - coordinate system. If we now multiply the index i by delta x we have the image coordinate x of the centre of pixel g (i,j). Similarly multiplication of the index j by delta y gives the image coordinate y. The traditional measurement of image coordinates is thus replaced in digital Photogrammetry by identification of pixels and this identification is, as far as possible, automatic ( refer to figure 4 ).

Photogrammetric restitutions also obviously require knowledge of the inner orientation. If the pixels are suitably small, it suffices to know which pixel contains the principal point. This thought leads us naturally to an extension whereby the indices i and j

110

can be interpreted as a image coordinate x and y. In this case, provided the pixels are square, the principal distance c can be introduced in units of delta x (= delta y). The restitution equations remain valid despite the unusual scale in the image and the image space. THE CREATION OF DIGITAL IMAGES:

If the same measuring accuracy as in analytical photogrammetry is to be reached

are even exceeded, the pixel size in the plane of the measuring camera must be as small as a few micrometres. If high accuracy is not the primary need, but rather one of the other advantages of the digital Photogrammetry is sought, a significantly larger pixel size

Y0 Y

0

X

Fig. 4 Digital image co-ordinate system suffices. A digital orthophoto is a typical, interesting product of digitawhich can be produced with significantly larger pixel sizes. If the time bimage and creation of end product is extremely short, digital PhotogramThis so called real time Photogrammetry, with large number of pixels, huand intensive data processing, is indeed possible today, but the resolvingdigital cameras falls far short of that of photographic cameras.

∆X

∆Y

X

l Photogrammetry, etween making the metry is essential. ge volumes of data power of present

111

Depending upon the job to be done, another method of creating digital images must therefore be selected. AUTOMATIC MEASUREMENT OF RESEAU AND FIDUCIAL MARKS:

If the photographs have been taken with a (large format) digital metric camera, we

know the principal distance and position of principal point. The inner orientation of the digital images is therefore, immediately available for the restitution.

If on the other hand the original image was made in metric photographic camera and the digital image is created by scanning of this photograph, restitution begins with the determination of the positions of the individual fiducial marks. This process should also be automated. The same task occurs when partial digital images are to be assembled numerically into the full image by means of a reseau.

The task is, therefore to find the position of geometric figure - we limit discussion here to a cross - in a digital image. We usually know the approximate position. The area around the approximate position is called search matrix or matrix of interest. The cross is called the target matrix. We also speak a template that shows 5 x 5 target matrix together with a 12 x 12 search matrix. For the sake of simplicity the densities are limited to the range 1 to 9 (refer to figure 5).

Fig.5: Reseau and fiducial marks in digital image

The cross in the search matrix obviously lies in the position i = 8 and j = 7. An automatic search for this position will be made harder by the facts that: • On the one hand, the search matrix is noisy, i.e., the densities are effected by random

errors, and • On the other hand because of the finite size of the scanning sensors the densities are

smeared, i.e., at the edges mixed pixels occur.

112

The solution of this correlation task is demonstrated below by an example, one dimensional for the sake of simplicity (refer to figure 6).

Target area

Search area

Fig. 6: Measurement of fiducial marks in digital image We determine the required position of the target area in the search area or area of interest, by means of correlation computations. A measure of the correlation is the correlation coefficient r, computed from the standard deviations S1 and S2 of the densities g1 and g2 in both areas and from the covariance S12 between the densities in both areas, as follows: S12 sigma of[(g1 - g11) x (g2 - g22)] r = --------- = ------------------------------------------------------------------------ S1 - S2 sqrt[sigma of{sqr(g1 - g11)} x sigma of{sqr(g2- g22)}] where g11,g22 = Arithmetic means of the densities of the target area and densities in corresponding section of the search area. We compute the correlation coefficient r for all possible positions of the target area in the search area and that position with the greatest value of r is the required position. ORIENTATION OF DIGITAL PHOTOGRAMMETRIC IMAGES:

The orientation of digital metric images can be directly compared with the numerical

orientation of metric photographs. Nevertheless, the result of orientation of the digital

113

metric image goes beyond mere determination of the six elements of the outer orientation of the metric photograph.

The orientation of stereo pairs or a bundle block adjustment has already been discussed in previous topic. For natural control points target areas must be created for automatic positioning in the digital images to be possible. The positions of the unsignalised tie points can also basically be found by automatic procedures. In this case, the target matrix is an extract from one digital image and the search matrix is a significantly larger extract from the other digital image.

After the orientation elements of two overlapping images have been determined, whether for a relative orientation or a complete absolute orientation, the conditions exist for the creation of normal-case images (normalised images). These corresponds to the normal case of metric photographs. Such normalized digital images therefore, play an important role in digital Photogrammetry, since the same conditions of homologous image elements apply as in human vision. The task of computer vision therefore imitates that of human vision. The two normalized images display only horizontal parallaxes and no vertical parallaxes. The correlation task is therefore now only one dimensional, since homologous points have the same x coordinates.

The mathematical relationship between the image coordinates x, y of the one of the original images and the image coordinates x', y’ of the corresponding normalized image can be derived from the collinearity conditions. - Spatial position of object space is taken into account - the negative principal distance f is substituted for (z-zo) and - Xo and Yo are set to zero. we have then x = -f (M11x'+M21y'-M31f')/(M13x'+m23y'-M33f') ...Eq (5.3.1) y = -f (M12x'+M22y'-M32f')/(M13x'+M23y'-M33f') A solution of eqs (5.3.1) for the image coordinates x' and y' is then: x' = -f' (M11x+M12y-M13f)/(M31x+M32y-M33f ) ...Eq (5.3.2) y' = -f'(M21x+M22y-M23f)/(M31x+m32y-M33f ) (refer to figure 7)

We have therefore now established the conditions for converting (resampling) the original digital images into normalised images. We define - in analogy to the procedure for digital orthophotos - the new image matrix in the normalized image (equ. 5.3.2). We select a principal distance f', somewhat larger than the principal distance f of the original images so as not to lose any pixels from the original images. We assume that the pixel size is the same in the original and normalized images. Given the coordinates x',y',c'of a pixel in the normalized image, we can apply equation (5.3.1) to find the corresponding position in the

114

original image. The required density must then be interpolated from the neighbouring pixels. The result of this resampling is the pair of normalized images.

The complete stereo model formed by a pair of normalized images can now be correlated by one-dimensional correlation. Corresponding image points in two normalized images lie as in (figure 8) in two lines with the same y coordinates (y1 = y2).

If we require only a few points rather than a complete stereo model we can eliminate the time consuming effort of resampling the original images and adopt an approach based on epipolar-ray geometry. This approach permits one-dimensional correlation even in the original images. An epipolar ray is formed by the intersection of the plane and epipolar plane, which is a plane passing through the two projection centres O1 and O2 and an object point P. All epipolar rays of the same image intersect in epipolar point K, which is the intersection straight line passing through the two projection centres with the image plane. An epipolar point is therefore the image of the other projection centre(the epipolar point of a normalized image lies at infinity).

The corresponding epipolar rays can be found in two-dimensional original images we can then apply a one-dimensional correlation along them. The two fundamental points defining the epipolar rays are the epipolar points K1 and K2. For e.g. the image coordinates of K1 can be found by inserting coordinates of the projection centre O2 in the collinearity equations. We can then obtain x(k) = -c (r11)/( r13) y(k) = -c (r12)/(r13) AUTOMATED PHOTOGRAMMETRIC POINT MEASUREMENT

We start with the assumption that the object points are signalized, though very well

defined natural points can also be used. The measurement of the image coordinates of such points in digital photogrammetric images can be very largely automated, for which purpose we must define target matrices for the individual signals. The positions of these signals in the digital images are then found in the same way as shown above for reseau crosses and fiducial marks.

There are two qualifications to this statement, however firstly, the signals must lie more or less in one-object plane and secondly, the image plane must be roughly parallel to this object plane. Under these conditions, applying principally to aerial photogrammetry, the signals are then about the same size and are not significantly deformed.

Until now we have left open the question of how the areas of interest in the digital image are found. The following methods may be used. For e.g.: - approximate positioning by an operator on a monitor screen

115

- Predefinition of approximate object coordinates and of the approximate elements of outer orientation of the individual images, followed by a central projection of the approximate object coordinates into the digital images.

C’ Y’

O Y

Fig. 7: Orientation of digital image

P’

P

P

P

X

x

X’

116

Z

Y X2

Y1 b

K1 X 2’

Fig. 8: Stereo model from digital image - When all required image points have been automatically identified and accurately

measured, a bundle block adjustment can be applied. Before beginning this however, the image coordinates must be refined by the known methods, depending upon the origin of the digital images. The coordinate refinement of the digital images should also include corrections for differences between the real pixel positions and the theoretical grid positions.

AUTOMATED SURFACE MODELLING

In a process to reconstruct the surface of the terrain by full automatic means,

digital image correlation has been proving very effective. In fact this is an area where digital photogrammetry has its strength . There are various correlation techniques. Some of them are frequently used techniques are given below. One-dimensional correlation

In this case correlation along particular direction is applied in the two normalized

images or by means of epipolar geometry in the two images. Correlation by distinct edges of features

Since the basic method of human correlation is based on first identifying identical

edges and then looking for correlation, this concepts also could be adopted for computer-

117

aided correlation in addition to the density correlation. This is a more complex method of matching which involves:

1. Edge enhancement and extraction by forming an image density. This is same as the convolution operation of image processing.

2. Finding corresponding points in two images by using comparison operations. 3. Computation of object space X,Y,Z coordinates based on the orientation

elements. Hierarchical multilevel correlation:

This is also known as a multiresolution matching by forming image pyramids of

different level. An image with finer pixels is progressively reduced to an equivalent image of coarser resolution by either elimination of pixels, averaging, applying higher order interpolation or by convolution operators. The correlation starts at the coarser level to subsequently reach at the finer level. Vertical line locus (VLL): This method of correlation is carried out on an oriented model. We begin by specifying X,Y coordinates of the required points in a regular grid. Then a series of equidistant planes perpendicular to each vertical passing through the X,Y points are established. By using collinearity equations to correspondingly Z coordinates. We can find homologous windows on each image and then correlation can be found by using correlation coefficients for each pair of windows. The maximum correlation defines the required X,Y,Z points. DIGITAL TERRAIN MODELLING

Digital terrain modelling is a particular form of surface modeling using computers,

which deals with the specific problems of representing the surface of the earth. The set of discrete digital coordinates in XYZ representing the terrain in its best possible form intended for specific application is called a digital terrain model. Several other terms are also used to describe essentially the same process such as digital elevation model (DEM), digital height model (DHM), digital terrain elevation model etc. There are various applications and generation methods of DTM. However in this lecture we will concentrate on photogrammetric methods of DTM generation and it's application in survey and mapping GIS and terrain visualization. PHOTOGRAMMETRIC METHODS :

The photogrammetric methods can be classified into four different categories;

manual, automatic by electronic correlator, semi automatic using analytical stereo plotter and fully automatic by digital image matching (auto correlation). At present the most widely used DEM extraction is carried out by using semi automatic analytical stereo plotters. Fully

118

automatic DEM extraction by digital image auto correlation is the most recent technique being selectively applied for certain applications using digital photogrammetric workstations. CONCEPTS:

Accurately modeling a solid object or a surface of regular nature is possible by

simple mathematical / geometric functions. How about the surface of the earth, which is not governed by any mathematical functions ? Only possibility is that we can be adaptive to a greater extent by choosing effective methods of sampling representing the smallest change of gradient and later on fitting a mathematical surface. However, one to one reconstruction of the terrain is not possible.

Optimum number of mass points covering the terrain is collected, along with the morphological information such as break lines, form lines, break points and cutout areas, which are processed for further densification by interpolation. MEASUREMENT PATTERNS (Semi Automatic using stereo plotters) :

In analytical plotters, automatic driving to the predefined points is possible using

computer programmes, which control the plate processors. The required terrain elevation can be derived by any one or combination of several sampling patterns. Systematic or regular sampling :

Automatic driving to the predefined points is useful for this type of sampling in

order to generate mass points on a regular grid. This is the latest method possible in semiautomatic plotters for data capture. However, this has a shortcomings of non-adaptability and optimization of number of points to be sampled. In some cases sampling distance is not appropriately chosen there could be a greater amount of redundancy and in other hand it may not be sufficient for a particular area in depicting the terrain morphology. This is more suitable for a terrain of uniform sloping. Progressive sampling :

The shortcomings of the above grid-based method is overcome by progressive

sampling (the idea originally proposed by Makarivic of the ITC, The Netherlands). The basis of the method is that one starts with a widely spread (low resolution) grid which will give a good general coverage of height control and on then a progressive increase in the sampling is done based on a threshold computed on the second difference on line to allow measurement of required number of points for densification, there by improving the terrain adaptability The second difference of height is calculated in both column and row directions to test the need for progressive densification. The mathematical expression for along row densification:

119

Delta (h) = hi+1 - 2hi + hi-1 If delta (h) > threshold then density. If delta (hij) < threshold then go ahead without densification. This process continues progressively patch wise. The threshold can be determined keeping in mind the ultimate fidelity in modeling aimed at and the terrain type. RANDOM SAMPLING:

Random sampling method is a selective mode of data collection where the operator

decides which is the salient terrain points and lines need to be sampled. It is highly subjective method, which depend on the operator's skill in appreciating the terrain. COMPOSITE SAMPLING:

This is a combination of regular and selective sampling. This approach is best suited

for a mixed terrain type. The basic grid pattern is supplemented by measurements made at hilltops, along break lines and streams. TERRAIN MODELLING

Sampled data alone is not sufficient to model the terrain. Hence further justification of the sampled data is done for more closer data points by adapting various interpolation techniques. Interpolation techniques are many, such as linear, moving average, polynomial etc. However in a broader level the modeling techniques can be divided into two classes. 1. Grid based terrain modeling 2. Triangular based terrain modeling Grid based terrain modeling:

It is best suited for automated systems, for example in photogrammetric applications the data capturing regular grids collected by photogrammetric means, however can be mixed with randomly distributed selective samples by finally further interpolating the random points to a regular grid. Here the computational approach is relatively simpler. Usually the following interpolation methods are distinguished. I. Point wise Interpolation of specific neighbouring points II. Global method fitting a single three dimensional surface defined by a higher order

polynomial through all the measured points Triangle-based terrain modeling:

This is a method where every data point measured is directly involved in computation

since they form vertices of the triangles used to model the terrain. Irregularly collected points can be more efficiently modeled by this method. Triangular irregular network(TIN) is

120

the terminology most frequently used to describe this method. One advantage of this method lies in its relatively easy way of incorporating break lines, fault lines etc. APPLICATIONS OF DEM:

DEM's have numerous applications however we will discuss here its application in

information technology: i. Contour generation for topographical mapping ii. Profiles, slopes for civil engineering applications iii. Highway alignments on hilly terrain by giving volume information for cut and fill

computation of earth work etc., iv. Perspective views for resource surveys and landscaping related application. v. Fly through perspectives for real time navigation aids. DIGITAL ORTHOPHOTOS: Introduction:

A line map, derived from aerial photographs is some times unsatisfactory for

undeveloped regions. Archaeologists, Soil Scientists, Foresters, Agriculturalists, Geographers, Geologists, Planners and Ecologists often do not find the details important for them in a line map. A photo which is equivalent of a map i.e. an orthographic projection of a ground is a better solution for them such photograph are called "ORTHOPHOTO". The process of generating orthophotos is known as orthophotography. This involves differential rectification i.e. rectification by small parts.

A map which shows the contents of the aerial photograph (photomap) is a better solution for them. This photomap is called as "ORTHOPHOTO", which is a photograph showing images of objects in their true orthographic positions. Therefore these are geometrically equivalent to conventional line and symbol planimetric map.

Because they are planimetrically correct, orthophotos can be used as maps for making direct measurements of distances, angles, positions and areas without making corrections for image displacements.

121

Orthophoto Production

Deformations:

Perspective photos have the problem of image displacements due to photographic

tilt and relief. Tilt displacement exists in any photo if at the instant of exposure, the photo plane is tilted w.r.t. the datum plane. Rectification eliminates the effects of tilt and yields an equivalent vertical photo. There exist a scale variations on the photo as a result of image displacement due to change in relief. These above deformations are removed with different methods described below. The resultant output is an orthophoto. Although relief displacement due to variable terrain are removed, a shortcoming of orthophotos is that relief displacements of vertical surfaces such as walls of building cannot be removed. Orthophoto instruments are broadly three types: 1. Production of image by direction optical projection (real time). 2. production of image by electronic means. 3. production of image by analytical means. 4. production of orthophoto by digital image processing.

The essential feature of the production of the digital orthophotos lies in the transformation of the image matrix in the camera coordinate system into an image matrix in the X Y plane of the ground coordinate system. The production of a digital orthophoto begins with the definition of the required image matrix in the X Y plane of the ground coordinate system, followed by a transformation of the centres of these elements into the camera coordinate system.

122

For this transformation we also need Z coordinates of the points in the X Y grid. These can result from a very close mesh of grid points measured in an analytical stereo plotter. Digital terrain models also provides the Z coordinates of a close mesh of XY points.

If we assume that the inner and outer orientations of the original digital image to be used for the orthophoto are known, the centres of the pixels in the ground system can then be transformed into the original digital image by the equations of central projection. Corrections for effects such as lens distortion and refraction can be applied in this process.

We come now to the still open question- How to assign densities in the original image to the transformed central points. The best-known assignment is that of the nearest neighbour. In this process the density of that pixel is adopted whose centre is closest to the transformed point. This method requires little computing time. A disadvantage is, however, that in the worst-case picture element are shifted by upto half a pixel. Relative displacement of upto one pixel can then occur in lines in the digital orthophoto. In order to ensure that no pixel in the original image is lost, the number of pixels in the digital orthophoto should be chosen to be significantly higher than in the original image.

An interesting alternative to the nearest neighbour method is the bilinear transformation. In this method, the density is determined by bilinear transformation from the four neighbours. The four densities g1, g2 ,g3 and g4 of the original image define a hyperbolic paraboloid on square of side delta(D), if we assume linear interpolation parallel to the coordinate axes. The density g(x',y') in position (x',y') can be computed from the following equation: g(x',y') = [1-x'/D - y'/D + (x'.y')/sqr(D)].g1 + [x'/D - (x'.y')/sqr(D)].g2 + [y'/D - (x'.y')/sqr(D)].g3 + [(x'.y')/sqr(D)].g4 A bilinear interpolation involves more effort than does the nearest neighbour method, but has the advantage that there are no breaks in lines. The original image contrast is however slightly reduced. If this reduction is to be avoided, a higher order interpolation must be used.

The summary below shows some of the advantages of digitally produced orthophotos compared with those produced photographically: • The geometric accuracy is basically higher since a very close mesh of points is used to

approximate the ground surface. • Image content can be modified quite simply by contrast manipulation of the densities

and colours. • An elegant matching of densities at the edges of neighbouring images in an orthophoto

mosaic can be achieved.

123

• Further improvements, such as edge enhancement, can be introduced by approximate filtering.

• The digital orthophoto can be stored as a level of information in a geographic

information system. • Digital orthophotos can be analyzed by the methods of multispectral classification,

image segmenting, pattern recognition, etc. DIGITAL PHOTOGRAMMETRIC WORKSTATIONS:

Software for the solution of the photogrammetric task based on digital images can, in

principle be installed in any digital computer. A relatively wide range of peripheral devices is necessary, however. The minimum requirements are : - Data capture unit to accept the data from digitized photograph, CCD cameras, etc. - Graphic screen with 640 x 480 pixels, better 1024 x 768 pixels. - 8 bit resolution, better 24 bit (8 bits for each of the primary colours red, green and

blue). - Processors speed 12.5 MIPS, better 25 MIPS. - 4 (better 8) MBytes main storage. - 650 MBytes mass storage, better two mass storage units.

Powerful PCs and so-called workstations can satisfy these requirements and computer systems tailored to the wide field of digital image processing are available today. More and more frequently additional printed circuit boards are being offered, fitted with a fast processors for suitable for such tasks as image convolution. Some of these systems will in future also contain software for digital Photogrammetry.

Similarly independent computer systems for geographic information system have been developed and the union of GIS and digital Photogrammetry will become even stronger in the foreseeable future. If digital orthophotos for e.g. are incorporated has background information in GIS, many tasks of digital Photogrammetry can be solved in such a workstation.

Some manufacturers are also offering independent digital photogrammetric workstations, incorporating three-dimensional observations of the digital stereo pairs.

124

How 3D stereo viewing is accomplished in a computer? Four different methods of separating the two digital images are in use at present: - Display of two images on full screen with observation through a mirror stereoscope. - Display of two colour composite images on the full screen with observation through

complementary colour spectacles ( Anaglyph process ). - Alternate display of the two images on the full screen, at a frequency about 50Hz and

observation through spectacles alternately passing and blocking light. A control is needed between the screen and spectacles to ensure synchronization mechanical versions of such spectacles have today been replaced by liquid-crystal shutter glasses.

- Alternating generation of the two images and synchronized display on a polarized

screen. The operator observes the screen through correspondingly polarized spectacles to achieve stereoscopy. The alternation between vision in the left and the right eye is also have achieved by liquid crystal spectacles.

Comparison of Analytical and Digital Photogrammetric systems:

Analytical instruments came into existence since 1970 and is being operated

presently. The soft copy or digital photogrammetry systems launched in 1990's. The Transition from analog/analytical equipment to soft copy photogrammetric system removes the need for specialized equipment. The major source of imagery for softcopy photogrammetry remains aerial photograph. The need exists therefore to convert the photographs into a pixel array using a film scanner; and the orthophoto need to be recorded on to film, creating a need for film writing equipment. There are scanners and film recorders are available at a reasonable prices.

125

In Digital photogrammetric system, the image stage system is the image storage, retrieval, and display memory systems, which are used to bring the digital image in to workstation's stereo viewing system. The Image Station stereo display software provides the image stage system. The image Storage is a typically of several Gigabytes magnetic disk or optionally an erasable optical disk.

A real time loop program in digital photogrammetric workstation collects user inputs, converts them to movements in object space, updates the object space position, transforms the object space to camera space with image refinements, transforms from camera space to pixel space, and from pixel space to window coordinates. Then either cursor is moved or the image is roamed. This whole thing is integrated into software. Stereo display system :

The image display system provides for the stereo display of imagery. It roughly corresponds to the stereo viewer portion of analytical stereo plotter including optics and stage system. With an analytical plotter, the stage move the film so the desired area is presented in the optical path for stereo viewing. Left and right optical paths deliver images to the left and right eyes through binocular eyepieces allowing stereo viewing. In DPS, there are several techniques available. 1. Anaglyph: One image is displayed in red and other in blue. Glasses with red and blue

filters are worn to provide the stereo effect. 2. Optical: The images are displayed on a split screen or in two CRTs and are viewed

through binocular optics similar to analytical plotter. 3. Polarizing screens and glasses 4. Image shuttering techniques 5. Lenticular viewing screens etc.. Graphic collection/edit system:

Many analytical plotters have had interactive graphics systems added to allow digital data capture. An interactive computer graphics system provides the basis for the capture/edit system of the DPS.

In a simple means, A Digital system is distinguished by number of points from

conventional photogrammetric instruments - No high precision optical-mechanical parts - Robust measurement system, no wear and tear

126

- No instrument calibration, no manual image handling - Combination of automatic and operator controlled processing - Data acquisition, processing, editing, storage, and administration in a single system - Automatic DEM generation by image correlation. - Perspective image viewing

- Fly through by displaying continuous perspective views at a high frequency

127

PRINCIPLES OF CARTOGRAPHY & MAP MAKING INTRODUCTION:

With the boom in human population and increased complexity of modern life coupled with attendant pressures and need to utilize optimally the limited resources, it has become indispensable to carry out detailed studies of the physical and social environment encompassing all fields of human activities. Scientists drawn from various spheres of human ventures e.g. geographers, planners, historians, economists, agriculturists, geologists, engineers etc., find map an indispensable tool and aid to carry on their pursuits. Maps, as such, may be: General maps: portraying the spatial association of a selection of diverse geographical phenomenon e.g. roads, settlements, boundaries, watercourses, elevations, coastline etc. OR Thematic Maps: portraying spatial variations of a single phenomenon or relationship between phenomenon e.g. diversity of soils, bedrock geology, population density, climate transportation etc.

Cartography can be briefly described as the art, science and technology of making maps of the earth or other celestial bodies to show the spatial relationships. Cartography is usually thought to consist of two classes of operations:

- Preparation of a variety of general maps used for basic reference and operational purposes. It includes larger scale topographical maps, hydrographic charts and aeronautical charts.

- Preparation of maps used for general reference and educational

purposes. It includes small-scale thematic maps, atlas maps, road maps and maps to accompany the written text in books.

128

Cartography can be classified into two categories:

- One type of cartography works primarily from data obtained by the field or hydrographic survey, or by satellite and photogrammetric methods.

- Other type of cartography includes thematic cartography,

draws on the basic work of the first group and pertains to communication of general information with effective graphic delineation of relationships, generalization and geographic concepts. The specific subject matter may be drawn from history, economics, urban planning, rural sociology, engineering and physical/social sciences.

METHODS IN CARTOGRAPHY: Conventional or Analog Method:

Preparation of maps usually involves the following processes:

- Choice of appropriate projection and coordinate system for plotting of spatial data.

- Compilation of data from existing documents supplemented with

field surveys or ground verification of details.

- Cartographic activities viz., generalization of data, choice of symbols, map-layout and design, and fair drawing.

- Map reproduction.

Compilation of Data from Aerial Photographs:

As early as 1840, it was suggested that photographs be used for the

purpose of mapping. By 1915, cameras specially designed for aerial photography were in use and during 1930s, there was extensive use of aerial

129

survey. Ever since, the aerial photographs (specially black & white) and photogrammetric methods have been reliable tools to furnish planimetric and hypsometric positions for map preparation. With limited ground control, aerial photos can provide accurate maps for areas where it is difficult to conduct extensive field surveys because of terrain or climatic conditions.

The aerial photographs taken over areas of varied relief suffer from 'Scale Variation' and relief displacement. In order to make distance and angle measurements on these photographs, compensation of relief effects in the photos is required necessitating their geo-referencing. The amount of displacement changes directly with the vertical departure from a chosen datum and the distance from the principal point, and inversely with the height of the camera. The displacement of objects on the aerial photographs produces parallax, i.e. apparent change in position of an object because of a change in the point of observation. On a photograph with no tilt, the parallax is a linear element used for determination of elevation. Densification of Control By Aero-triangulation: (a) Graphical Method

Since azimuths from the principal point are correct to any point on

the photographs, graphical triangulation was performed directly on photographs for extension of control over large areas by using radial line plotting e.g., Hand Template Method, Slotted Template Method or Radial Arm Template Method. In radial line plotting, each point is relocated by the amount of its displacement and all points are then located at common point. The actual scale of plot depends on the distance between the principal points of two consecutive & overlapping photographs. Usually a radial line plot is prepared at a pre-determined scale by using control points provided in the field, or read from reasonably accurate map. Preferably, three control points are plotted in the area of overlap of the first two photographs. (b) Optical Method (Bundle Ray Adjustment)

As technical improvement over the radial line plotting, the optical

method or Bundle Ray Adjustment method involves:

130

- Creation of stereo-models with the help of aerial photos. - Integration and adjustment of strips of stereo-models. OR - Simultaneous adjustment of all stereo models using IMT method.

Creation of stereo-model essentially involves the reconstruction of camera-geometry used at the time of photography such that the bundle of rays intersect in space to present ground image. The above adjustment requires ground control points which are usually established by conventional methods of ground survey viz., triangulation, trilateration, traversing and leveling. Currently, the establishment of ground control is aided by the use of GPS procedures. The accurate ground control is essential to virtually all-photogrammetric operations, because photogrammetric measurements can only be as reliable as the ground control on which they are based. These ground control points must clearly be identifiable both on the ground and on the photography being used. Stereo plotters (Mechanical/Opto-mechanical):

These are precision instruments designed for preparation of

topographic maps. Herein, two projectors are used that can be adjusted in their position and angular orientation to duplicate the exact relative position and orientation of the aerial camera at the instance the two photos of stereo-pair were exposed. Similarly, the base distance between exposures and differences is flying heights are simulated by adjusting the relative position of the projectors. A stereo-plotter is made of three basic components:

- A projection system (to create terrain model)

- A viewing system (to view the model stereoscopically)

131

- A measuring and tracing system (for measuring elevations in the model and tracing features onto a map sheet.

The instrument uses reduced size diapositives. Fair-drawing (Inking/Scribing) From Ground-verified Manuscripts/sheets:

Verified plane-table sheet or air survey sections are first

photographed on the scale of fair mapping and mosaic on glass, paper or zinc sheet using black prints, bromide prints or film negatives. This process of mosaicing involves:

- Projection of map sheet on drawing paper at the scale of fair drawing.

- Plotting of trigonometrical stations and permanent traverse

stations on the projection.

- Trimming of prints or film negatives to about 6mm outside graticule or common edge and fitting them to the projection by matching the graticule lines where they leave the print sets. Where prints do not fit, they are cut into sufficient number of sections to reduce the discrepancy between sections to not more than about 0.3 mm. All the outer sections along graticule lines are pasted down first and the inner sections are then fitted to distribute the error equally. The above mosaic prints are then used for preparation of combined negatives by photography which are further used to obtain:

- Black or blue prints on good quality map Litho, bank-post or

tracing paper for accessory work, i.e. preparation of guides etc.

- The blue prints on drawing paper (at scale 50% larger than the scale of actual survey) as obtained above are checked for correct dimensions and are used to prepare -

* Name Original (including marginal information)

* Contour original * Outline or Black Original * Red Original * Blue Original * Green Tree and Tint Original

132

These originals (drawn with pen and ink) are further used in preparation

of printing plates for different process colours (cyan, magenta, yellow and black and subsequent offset printing of maps. The drawing work done on paper with ink suffers from: -

- Regular/irregular distortion of the dimensions of sheet due to warping of paper caused by temperature and humidity.

- Fading of ink over time making the original unfit for photography. - Inconsistency in drawing work by different draughtsman. - Wear and tear of paper on which the fair drawing is done.

With the introduction of Mylar based scribe-sheets and peel-coats, which ensure reasonable dimensional stability, scribing technique, superseded the conventional fair drawing with pen & ink on paper. As a result, the diazo-prints of mosaic plane-table sheet or air-survey sections are directly obtained on the scribe-coats and peel-coats, at the scale of survey and originals are drawn/etched with the help of scribing tools. The scribed originals and peel-coat masks act as negatives and can be used as such for photography to prepare plates for offset printing. The use of scribe-coat and peel-coat sheets reasonably eliminates the possibility of dimensional distortion of sheets and offers consistency and neatness in drawing work. Digital Method:

The preservation of field data in form of hard copy of maps, field

sheets, printing plates or film negatives has been a source anxiety due to their quick perishability or distortions, besides unmanageable space required to store them. The advent of computers and digital storage devices, stimulated a chain of revolutionary changes in the cartographic procedures viz. projection, compilation of data, drawing, map reproduction and data

133

storage. Thus, the concept of Computer Assisted Cartography (CAC) emerged in 1970s. Projection:

Until the use of computers, construction of projection was done

manually. Now, with commonly available software, any projection based on a set of equations or table of values, can be produced by computer plotter, or can be displayed on screen as a soft copy. Projection is a process of producing all or part of a curved/round body on a flat sheet. Since projection cannot be done without geometrical distortion, the cartographer chooses a suitable developable surface for projection (e.g. plane, cylindrical or conical) and a suitable characteristic (equal-area or equivalence, correct shape or conformality, correct scale or equi-distance, correct direction or azimuth) which is to be shown accurately at the expense of others. He also decides upon the types of aspect (normal, traverse or oblique) and contact (tangential or secant) of the projection surface with respect to the curved/round body. Analytical Stereo plotter:

Against the mechanical or opto-mechanical stereo-plotter, analytical

plotter operates through the formation of a mathematical model of the terrain imaged by a stereo-pair. This is done linking a comparator type viewing and measuring system to a digital computer and as such, the system becomes extremely accurate and versatile. The system's computer can be programmed to handle any type of photography (e.g., oblique or panoramic photos), and to correct for complex combination of image distortions. The operator simply feeds the camera focal length and other calibration data into the computer. Then under cursor control, the coordinates of the fiducial marks and some ground control points are measured and the computer performs complete orientation of the stereo model. Positions in the model may be mathematically transformed into ground coordinates and elevations. Such system allows simultaneous viewing of the stereo-model and digitized line work. A graphics monitor is used for reviewing and editing the digitized data. There are software’s which permit menu-driven coding of map-features and digital elevation model production.

134

Soft copy Photogrammetric Workstations: The soft copy system uses digital raster images rather than

photographs. This system extensively uses mathematical modeling and incorporates not only the functionally of analytical plotters but also permit the integration of various photogrammetric tasks into computer-based environment, e.g. automated generation of digital terrain models, computation of digital orthophotos (for subsequent output on a raster plotter), capture of data for direct entry into GIS, etc. The system also provides linkages to image processing software making it amenable to the analysis of any kind of digital image data. The examples of above workstations are Leica DPW 710 and Sun Sparc Matra, IMD of Intergraph, PHODIS of Zeiss. Cartographic Production Line (Digital Method): Sources for Compilation of Data: a) Aerial photos to prepare stereo-plots/air survey sections used for

subsequent field verification. b) Digital raster image of aerial photos. c) Satellite imagery preprocessed and geo-referenced. The pre-processing

of imagery includes:

- Corrections for geometric distortions e.g. variation in altitude, attitude and velocity of sensor platforms, panoramic distortion, earth corrections, atmospheric refraction, relief displacement, non-linearity in the sweep of a Sensor's IFOV.

- Radiometric correction for Sun-elevation, earth-sun distance

and haze.

- Removal of noise, i.e. stripping or banding, line drop and bit errors (non-systematic variations in grey-levels from pixel to pixel).

135

d) Existing maps corrected for dimensional distortion of the paper base. The map may be manually digitized, or alternatively, scanned and vectorised in a semi-automatic mode, i.e. automatic vectorisation with interactive editing.

e) Digitization of field - verified P.T. sections or air survey sections,

manually or in semi- automatic mode. f) Data Editing:

Digital data generated through digitization of field records or

vectorisation of digital raster images followed by projection transformation, is used to create graphics for mapping, which involves following steps:- a) The features of various classes are transferred to different layers or

levels. For example, contours, vegetation, drainage or water-features, settlements, roads and railways, text etc. are stored in different layers. Even the details of a feature class are stored in different layers to afford GIS possibility. The structures of various layers are organized in such a way that one layer does not obscure the details of other layers, and correct topological relationships of the spatial features are preserved. For example, the layer of road should be above the layer of river or canal as per the ground reality, and so on.

b) According to the scale and purpose of map, map-layout is prepared and

designed showing:

- The position of legend, and marginal details i.e., title of the map, bar-scale, reference squares/graticule or grid values, magnetic declination, sheet Nos. & edition, administrative and sheet indices, special footnotes etc.

- Symbols with suitable size, colour and style

- Texts of suitable font size and style

- Dimensions of borders and neat lines

136

- Position of inset maps

- Type of map: island, bleeding edge or bounded. c) Generalization of details depends on the purpose of map, scale,

graphic limits and quality of data. First part of generalization i.e., conceptual generalization, involves symbolization and classification of features into distinct feature classes by grouping like features. The second part of generalization i.e. graphic generalization which is implemented at the time of drawing features on the computer, consists of elimination of unwanted details and graphic processes e.g., smoothing lines, deleting small details or combining those small individual features of one feature class which would merge or simply become invisible when reducing them unchanged to a smaller scale, exaggeration of details (e.g. roads) or displacement of details (e.g. features adjacent to the exaggerated features) to ensure legibility of map. Preferential order of displacement is:

- Area features e.g., forests - Buildings - Roads - Waterways - Railways

Of late, efforts have been made to switch from manual map generalization to automatic map generalization. The long-term goal of automated generalization is a 'Scale-less Mapping System'. Such a system contains one large dataset of mainly topographic information plus some attribute information that can be manipulated and generalized in whatever degree to produce maps on every desired scale. This would have some clear advantages over the present practice of digitizing and storing at each scale independently. While automation of conceptual generalization is easier to achieve, the graphic generalization which is determined by subjectivity, and intuition based on long-term experience and feeling is quite complicated for automation.

137

d) Symbols are designed for point, linear and area features specifying the size of symbol, thickness of line, colour and pattern of area in till. While designing symbol, the level of perception (i.e., nominal, ordinal or ratio/interval) is taken care of for choosing visual variables (e.g., position, orientation, form, size, colour, value and textural).

e) Various details are drawn according to the chosen map-design, and

edited to ensure the legibility and effective visual presentation of the map, using colour proofs of the map obtained from Laser printer which uses post-script file of the map.

f) The map file of the final corrected map is prepared for obtaining

colour separates on film (positive or negative) digitally or for automatic scribing of details of different layers and preparation of peel-coat masks. Map files so created may be converted to the standard data format and merged into topographical database or any other subject/application database.

g) Colour separates prepared digitally from the map file may be used to

prepare plates for offset printing. Alternatively, the scribed originals and peel-coat masks may be used to prepare combined film negatives or diapositives by contact photography for each process colour cyan, magenta, yellow and black, which may in turn, be used to prepare plates for offset printing. Of late, the digital offset printing techniques have been displayed at DRUPA '95 held in May, 1995 in Dusseldorf (Germany), which do not require printing plates for offset printing; instead, the map file is imported to the digitally driven offset printing machine (mono block) and printing of map is done automatically with the help of the magnetic images imprinted on the printing cylinders and magnetized ink. Presently this technique is limited to the printing on A3 size.

Data Exchange Formats for Cartographic Data:

In order to facilitate exchange of digital data among various user

agencies, a standard data format is evolved for storage/archival and

138

transfer of digital data. In India, the national data exchange format is DVD (Digital Vector Data) for all kinds of topographical data. Other formats in vogue elsewhere are:- - DIGEST (Digital Geographic Exchange Standard) used in NATO countries - NTF (National Transfer Format) used in the United Kingdom - SDTF (Spatial Data Transfer Format) and DLG used in the U.S.A. - EDIGEO used in France - Standard International format (ISO 8211)

139

COMPARISON OF CONVENTIONAL (ANALOG) AND DIGITAL METHODS OF CARTOGRAPHIC PROCESSES: Sl.No

. Features/process Analog Method Digital Method

1.

Projection

Manual, tedious

Easy

2.

Transfer of Data from one projection to another

Tedious

Quick and efficient

3.

Interpolation of data

Tedious

Easy and efficient

4.

Source of data compilation

Aerial Photos, Field Records

Aerial Photo, Digital image, Satellite imagery, Field records

5.

Fair-drawing scribing

Use of ink & pen and scribing tools

Use of digital methods of drawing techniques based on vectorised raster data or vector data

6.

Map Reproduction

Use of film negatives to prepare printing plates for offset printing

Direct preparation of colour separates on film which can be used to prepare printing plates for offset printing. Alternatively, Digital printing of maps

7.

Data storage and archieving

Hard copies of maps are

Digital files of map sheets can be stored on magnetic

140

perishapble and difficult to store

tapes, optical disks or magneto-optical disks for longer preservation.

DATABASE:

This is a collection of inter-related data stored together with

controlled redundancy to serve one or more applications in an optimal fashions. The data are stored so that they are independent of programs which use the data. A common and controlled approach is used in adding new data and modifying and retrieving existing data within the database. Database Characteristics: i) Data independence ii) Speedy handling of spontaneous information requests iii) Non redundancy iv) Versatility in representing relationships between data items v) Security protection vi) Real time accessibility Data Independence: Logical Independence:

Overall logical structure of the data may be changed without changing the application program.

Physical Independence:

Physical layout and organization of the data may be changed without changing either overall logical structure of data or application program.

Purpose of Data-base Management System:

The main purpose in moving to a database environment is to achieve faster, more flexible application development and low maintenance cost.

141

Some advantages of organizing data using DBMS data using DBMS may be listed as below:-

- Minimizing redundancy of data storage - Central control on access to data - Easy manipulation of data - Data integrity - Security of data - Making application programs independent of the form in which

data is stored Data Environment: Class Environment:

A data base management system is not used and there are separate files of data for most application software’s, VSAM, BDAM, DMS. This is almost obsolete. Class II Environment: It consists of Application Database: TOTAL, IMS, IDMS, IDS. Class III Environment:

It consists of Subject Databases, which are largely independent of specific application. Data are designed and stored independently of the function for which they are used. Examples of database are : IMS, IDMS, IDS, ADABAS. Data Dictionary:

The main functions of Data Dictionary are:

- To inform people about data - To help control the definition and representation of data - To indicate which programs are affected when changes are

made to data structures or representation Data Dictionary should preferably be independent of DBMS. The salient features of Data Dictionary are :

142

- Support definition of all types of data items, data groupings

and data associations

- Permits information to be stored about the data associations

- Gives support for file systems as well as database

- Supports definition of process entities e.g. systems, programs, modules, projects, transactions, etc.

- Supports attractively formatted, easily comprehensible reports

of all aspects of dictionary usage

- Automatically captures data in the existing program

- Displays dictionary information interactively

- Supports the definition of security levels and authorization details who can do what with the data

- Built-in-security so that the data in the dictionary cannot be

tampered

- Generates the control blocks and parameters used by DBMS programs

- Generates screen or report formats for high level language

features

- Enforces used by the programmers of data definitions that are in the dictionary

- Automatically converts data items to the same formats before

adding or otherwise manipulating them in combination. This implies a tight combination of dictionary and DBMS.

143

Data Model: Hierarchical Model:

It is based on tree-structure. Here, tree is composed of a hierarchy of elements called nodes. The uppermost level of hierarchy has only one node, called the root. With the exception of root, every node has one node related to it at a higher-level called parent. No element can have more than one parent. Each element can have one or more elements related to it at lower level called children. Elements at the end of the branches i.e. with no children, are called leaves. Thus, in the hierarchical model, each record in the hierarchy (except for those at the top) is associated with one record in the next higher level of hierarchy.

- Main features of hierarchical model are:

- Information is retrieved by traversing the Tree structure using the procedural query language.

- Easy to update the hierarchical system.

- Searches cannot be done on the attribute field.

- Data relationships are difficult to modify.

Examples of hierarchical database : IMS, DL/1, TOTAL, IMAGE, SYSTEM 2000. Network (CODASYL) Data Model:

Herein, the records are grouped into two-level hierarchies called sets. These sets can overlap to form networks, i.e. a record can be part of more than one set. Main features of Network Data Model are :

- Less redundant data storage than the hierarchical model

- Requirement and storage of more extensive linkage information

144

- Updating linkages is tedious and time consuming Relational Data Model:

All data are represented as records without pointed linkages. Instead, the records contain data items which allow the necessary associations to be made. This approach gives simple tabular structure of data. Basically, the relational model allows the use of powerful, set-oriented, associative expressions instead of the one-record-at-a-time primitive of more procedural models like the CODASYL model. In the relational databases, data are stored in tables, called relations. Each relation has a fixed number of columns called attributes and a (dynamic, time varying) number of rows called tuples. The number of attributes of a relation is called its grade; the number of types is called cardinality. The set of possible values for a given attribute is called its domain. The following operations can be performed on relations, each of which takes one or two relations as operands and produces one relation as result : Unary Operations : Selection, Projection Binary Operations : Union, Difference, Cartesian product, Natural Join, Semi- Join, Natural Semi-Join Examples of Relational Database : DATACOM, NOMAD, DBMS Main features of relational data model are :

- Flexible manipulation of data - Simple data organization - Difficult to maintain - Use of non-procedural language (e.g. SQL)

Spatial Data Model: There are two approaches to represent the spatial component of

geographic information : Vector Model:

145

The objects or conditions in the real world are represented by the points and lines (including polygons) that define their boundaries. Mainly, the vector model is used to represent distribution of objects in space or conditions that apply to an area feature. The vector model can further be sub-divided into the following two classes : Spaghetti Model:

In this model, a paper map may be translated line-for-line into a list of xy - coordinates. A point is encoded as a single xy-coordinate pair, and a line as a string of xy pairs. An area is represented by a polygon and is recorded as a closed loop of xy coordinates that define its boundary. A file of spatial data so constructed is a collection of coordinates with no inherent structure. Topological Model:

- It consists of construction of following tables :

- Polygon Topology Table : showing arcs of boundaries of each polygon.

- Node Topology Table : showing arcs which define the nodes.

- Arc Topology Table : showing the starting Node and End-node

of each arc along with polygons on the left and right of arc.

- Arc Coordinate Table : each arc is represented by one or more straight-line segments defined by a series of coordinates.

Main features of Vector Model are :

- More compact data structure - Efficient network analysis - Better suited to support graphics

Raster Model:

Space is sub-divided into cells (usually square in shape). The location of geographic objects or conditions is defined by row and column positions of the cells they occupy. Area covered by cell is the spatial resolution. Thus,

146

147

Raster Model is used to represent the spatial variability of phenomenon. Main features of Raster Model are :

- Simple data structure - Overlay operations easy (arithmetic (+, -, /, *), logical) - Efficient representation of spatial variability of phenomenon - Efficient manipulation and enhancement of digital images

Reference: 7.1 Robinson, A., Sale,R., Morrison, J., "Elements of Cartography" (1978). 7.2 Wolf, Paul R., "Elements of Photogrammetry" (1983). 7.3 Lillesand, Thomas M., Kiefer, Ralph W., "Remote Sensing and Image Interpretation" (1994). 7.4 I.T.C. Notes.