intensity transformations (chapter 3) cs474/674 – prof. bebis

52
Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Upload: marshall-fletcher

Post on 15-Jan-2016

238 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Intensity Transformations (Chapter 3)

CS474/674 – Prof. Bebis

Page 2: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Spatial Domain Methods

f(x,y)

g(x,y)

g(x,y)

f(x,y)

Point Processing

Area/Mask Processing

Page 3: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Point Processing Transformations

• Convert a given pixel value to a new pixel value based on some predefined function.

Page 4: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Identity Transformation

Page 5: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Negative Image

• O(r,c) = 255-I(r,c)

Page 6: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Contrast Stretching or Compression

• Stretch gray-level ranges where

we desire more information

(slope > 1).

• Compress gray-level ranges that

are of little interest

(0 < slope < 1).

Page 7: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Thresholding

• Special case of contrast compression

Page 8: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Intensity Level Slicing

• Highlight specific ranges of gray-levels only.

Same as doublethresholding!

Page 9: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Bit-level Slicing

• Highlighting the contribution made by a specific bit.• For pgm images, each pixel is represented by 8 bits.• Each bit-plane is a binary image

Page 10: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis
Page 11: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Logarithmic transformation

• Enhance details in the darker regions of an image at the expense of detail in brighter regions.

compress

stretch

Page 12: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Exponential transformation

• Reverse effect of that obtained using logarithmic mapping.

compress

stretch

Page 13: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Equalization

• A fully automatic gray-level stretching technique.

• Need to talk about image histograms first ...

Page 14: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Image Histograms

• An image histogram is a plot of the gray-level frequencies (i.e., the number of pixels in the image that have that gray level).

Page 15: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Image Histograms (cont’d)

• Divide frequencies by total number of pixels to represent as probabilities.

Page 16: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Image Histograms (cont’d)

Page 17: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Properties of Image Histograms

• Histograms clustered at the low end correspond to dark images.• Histograms clustered at the high end correspond to bright

images.

Page 18: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Properties of Image Histograms (cont’d)

• Histograms with small spread correspond to low contrast images (i.e., mostly dark, mostly bright, or mostly gray).

• Histograms with wide spread correspond to high contrast images.

Page 19: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Properties of Image Histograms (cont’d)Low contrast High contrast

Page 20: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Equalization

• The main idea is to redistribute the gray-level values uniformly.

Page 21: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Equalization (cont’d)

• In practice, the equalized histogram might not be completely flat.

Page 22: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Probability - Definitions

• Random experiment: an experiment whose result is not certain in advance (e.g., throwing a die)

• Outcome: the result of a random experiment

• Sample space: the set of all possible outcomes (e.g., {1,2,3,4,5,6})

• Event: a subset of the sample space (e.g., obtain an odd number in the experiment of throwing a die = {1,3,5})

Page 23: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Random Variables - Review

• A function that assigns a real number to random experiment outcomes (i.e., helps to reduce space of possible outcomes)

X(j)

X: # of heads

Page 24: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Random Variables - Example

• Consider the experiment of throwing a pair of dice• Define the r.v. X=“sum of dice”• X=x corresponds to the event

Page 25: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Probability density function

• The probability density function (pdf) is a real-valued function fX(x) describing the density of probability at each point in the sample space.

• In the discrete case, this is just a histogram!

Gaussian

Page 26: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Probability distribution function

• The integral of fX(x) defines the probability distribution function FX(x) (i.e., cumulative probability)

• In the discrete case, simply take the sum:

fX(x) FX(x)

GaussianfX(a)da

non-decreasing

Page 27: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Probability distribution function (cont’d)

Page 28: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Uniform Distribution

fX(x) FX(x)

[email protected]

Page 29: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Random Variable Transformations

• Suppose Y=T(X)– e.g., Y=X+1

• If we know fX(x), can we find fY(y)?

• Yes - it can be shown that:

Page 30: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Transformations of r.v. - Example

μ=0, σ=1

μ=1, σ=1

0

Page 31: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Special transformation!

1 1( ) ( )

1( ) [ ( ) ] [ ( ) ] 1

( )Y X Xx T y x T yX

dXf y f x f x

dY f x

Proof:

Page 32: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Equalization (cont’d)

The intensity levels can be viewed as a random variable in [0,1]

pr(r) ps(s)

Page 33: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Equalization (cont’d)

For PGM images:L=256 (graylevels)k=0,1,2, …, L-1 (possible graylevels)rk=k/(L-1) (normalized graylevel in [0, 1])

Page 34: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Equalization (cont’d)

then, de-normalize: sk x (L-1)

Page 35: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Equalization Example

3 bit 64 x 64 image

input histogram equalized histogram

Page 36: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Equalization Examples

equalized images and histogramsoriginal images and histograms

Page 37: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Specification (Matching)

• Histogram equalization yields a uniform pdf only.• What if we want to obtain a histogram other than uniform?

fR(r)

fR(a)da fZ(a)da

fZ(z)

onlydesiredhistogramis known

so, Q(r)=G-1(T(r))

Page 38: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Specification (cont’d)

• fS(s) and fV(v) are uniform• G(z) can be computed by specifying fZ(z) but I2 and I’2 are

unknown!• z=G-1(v) requires that v is a r.v. with uniform pdf• IDEA: use z=G-1(s) instead of z=G-1(v)

– s is a r.v. with uniform pdf• The desired transformation is

z=G-1(T(r))

fR(a)da fz(a)da

Page 39: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Specification (cont’d)

• Comments– We do not need to apply T( ) and G-1( ) separately!– Combine them: Q=G-1T, thus, z=Q(r)– Histogram specification assumes that we know G-1 (not

always easy to find).– G( ) must be in [0,1] and must be non-decreasing.

z=G-1(T(r))=Q(r)

Page 40: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Specification Example

3 bit 64 x 64 image

specified histogram

actual histogram

input histogram

Page 41: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Specification (cont’d)

• Histogram specification might yield superior results than histogram equalization.

results of histogram equalization

Page 42: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Specification (cont’d)results of histogram specification

G(z)

G-1(z)

Page 43: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Local Histogram Processing

• Histogram equalization/specification are global methods.– The intensity transformation is computed using pixels from

the entire image.

• Global transformations are not appropriate for enhancing little details in an image.– The number of pixels in these areas might be very small,

contributing very little to the computation of the transformation.

Page 44: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Local Histogram Processing

Idea: Define a transformation function based on the intensity

distribution in a neighborhood of every pixel in the image!

Page 45: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Local Histogram Processing (cont’d)

1. Define a neighborhood and move its center from pixel to pixel.

2. At each location, the histogram of the points in the neighborhood is computed. Obtain histogram equalization or histogram specification transformation.

3. Map the intensity of the pixel centered in the neighborhood.

4. Move to the next location and repeat the procedure.

Page 46: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Local Histogram Processing: Example

local histogramequalization3 x 3 neighborhood

global histogramequalization

Page 47: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Histogram Statistics

1

0

( ) ( ) ( )L

nn i i

i

r r m p r

1

0

( )L

i ii

m r p r

1 1

0 0

1( , )

M N

x y

f x yMN

12 2

20

( ) ( ) ( )L

i ii

r r m p r

1 1

2

0 0

1( , )

M N

x y

f x y mMN

n-th momentaround mean

Variance(2nd moment)

Mean(average intensity)

Page 48: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Example: Comparison of Standard Deviation Values

31.6 14.3 49.2

σ is useful for estimating image contrast!

Page 49: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Local Histogram Statistics

1

0

Local average intensity

( )

denotes a neighborhood

xy xy

L

s i s ii

xy

m r p r

s

12 2

0

Local variance

( ) ( )xy xy xy

L

s i s s ii

r m p r

Page 50: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Using Histogram Statistics for Image Enhancement

• Useful when parts of the image might contain hidden features.

Task: enhance darkareas without changingbright areas.

Idea: Find dark, low contrastareas using local statistics.

Page 51: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Using Histogram Statistics for Image Enhancement: Example

0 1 2

0 1 2

( , ), if and ( , )

( , ), otherwise

: global mean; : global standard deviation

0.4; 0.02; 0.4; 4

xy xys G G s G

G G

E f x y m k m k kg x y

f x y

m

k k k E

Page 52: Intensity Transformations (Chapter 3) CS474/674 – Prof. Bebis

Question

• Intensity operations can yield pixel values outside of the range [0 – 255].

• You should convert values back to the range [0 – 255] to ensure that the image is displayed properly.

• How would you find the following mapping?

[fmin – fmax] [ 0 – 255]