OPTIMAL TELEVISION SCANNING FORMAT FORCRT-DISPLAYS

Erwin B. Bellers, Ingrid E.J. Heynderickxy, Gerard de Haany, and Inge de WeerdyPhilips Research Laboratories, Briarcliff Manor, USA

yPhilips Research Laboratories, Eindhoven, The Netherlands

ABSTRACT

At the time of introduction of the television, the video for-mat was optimized given both economical and technologicalconstraints. The resulting video format is not necessarilyoptimal for the current display technology. Moreover, cur-rent consumer-level priced and state-of-the-art scan-rateconverters enable a spatio-temporal decoupling of the re-ceived video and the displayed video. This paper presentsthe results of a subjective assessment indicating the pre-ferred CRT-display format.

1 INTRODUCTION

Interlaced video with 525 or 625 lines and a 50 or 60 Hzpicture rate has been the television broadcast standard forquite some time. Modern bright television screens, how-ever, require a modified display format to prevent annoyinglarge area flickerand/orinterline flicker. Moreover, matrixdisplays require format conversion as they cannot directlycope with interlace. Finally, new image sources, such as theInternet, benefit from an increased spatial pixel density toimprove the legibility of displayed textual and graphical in-formation. As such, the optimal scanning or video formatrequirements may differ per application.

The techniques for high qualityvideo format conversionhave recently reached a price / performance ratio that en-ables application in the consumer domain [1, 2, 3]. As a re-sult, we can choose the displayednumber of scanning lines,the interlacefactor and thepicture rateat will. Given thisnew freedom, the question arises how to optimally choosea display format for the current applications. Increasing thenumber of scanning lines increases the vertical resolution.Modifying the interlaced scanning to the progressive for-mat eliminates any of the possible interlace artifacts likeline flickeringandline crawl. Finally, an increase in the re-fresh rate reduces or eliminates thelarge area flicker. Con-sequently, one might expect optimal performance by usingboth the highest number of scanning lines and at the high-est refresh rate possible. Obviously, there is a cost increaseassociated with such an increase in overall quality.

In order to have a fair optimization of the television dis-play format, i.e. comparing options with approximatelythe same cost1, we mayincreasethe number of scanninglines at the expense of a lowerrefresh rate,or a changein the interlace phase. Note that this recently obtained ad-ditional freedom enables a variety of pixel distributions inboth space and time, for every chosen pixel rate.

Section 2 focuses on the conducted experiments and se-lected environment. Section 3 presents the results of theexperiments. Finally, in Section 4 we draw our conclusions.

2 THE EXPERIMENTS

A video format is mainly characterized by the spatial reso-lution, temporal frequency, i.e. update frequency, and theinterlace phase. As explained above, a fair optimisationshould compare alternatives at equal cost, which impliesthat we have to look for the right balance from a viewer’spoint of view between the number of scanning lines, the re-fresh frequency and the interlace factor at a given pixel fre-quency. As far as we know, no objective metric yet is ableto reliably predict the perceived quality resulting from com-binations of these three video format characteristics. There-fore, subjective testing is the only available option for de-termining the optimal video format.

2.1 Video formats

From a viewer’s point of view the optimal video formatwill be the one that offers the best balance between verti-cal resolution, flickering and annoying interlace artefacts.In those cases that subjects have to balance various aspectsof one stimulus (i.e. a still image or a sequence), the paired-comparison technique is the preferred methodology. It im-plies that a subject is shown a pair of stimuli, where eachstimulus of the pair is displayed in a different video format.The subject is requested to indicate the stimulus that is mostpreferred. In this way, each subject has to compare for a set

1the same pixel and line frequency

VideoFormat

Conversion

input format

625(1:1)@50Hz

display format

e.g. 625(2:1)@50Hz 1250(2:1)@50Hz

833(2:1)@75Hz

625(2:1)@100Hz

compare

525(1:1)@60Hz

Figure 1: Decoupling of the input video format and the dis-play format.

of images with various content all mutual combinations ofthe different video formats in the study.

The disadvantage of subjective testing in general is thatsubjects are only able to assess a limited number of stimuli.Therefore, especially with the paired-comparison method-ology the number of video formats that can be evaluated ona selection of image material has to be restricted to about 5formats. The 5 formats we carefully selected on availabil-ity and applicability differ in the number of scanning lines,the interlace phase and the refresh rate, but all use the samepixel frequency (27 MHz2). These formats are (see alsoFigure 1):

• 50 Hz, progressive (1:1), 625 scanning lines (50p)• 60 Hz, progressive (1:1), 525 scanning lines (60p)• 50 Hz, interlaced (2:1), 1250 scanning lines (50i)• 75 Hz, interlaced (2:1), 833 scanning lines (75i)• 100 Hz, interlaced (2:1), 625 scanning lines (100i)

As indicated, all these formats have an intrinsic higherspatio-temporal resolution that the common standard defi-nition video format. As such these formats, therefore, po-tentially result into an improved appreciation of the standarddefinition video displayed nowadays, while at the same timethe larger sampling frequency has proven to be a viable op-tion.

2This is twice the pixel frequency for common interlaced video.

As a first format considered in the evaluation we in-creased the vertical number of lines to arrive at a progres-sive video signal indicated as 50p (for so-called ’50Hz-countries’). Similarly we selected 60p as a viable option.

Instead of displaying the 625 lines of the 50p formatwith the same interlace phase (because it is progressive), wecould also consider to display the same amount of lines perfield but with a different interlace phase (50i). As such, wecreate the opportunity to further increase the vertical resolu-tion by a factor of 2! As nice as this might seem, in practice,we have to take the typical interlace artifacts, especially ata refresh rate of 50Hz, into account. And as a result, weneed to balance the visibility of interlace artifacts versus thepotential double vertical resolution.

The visibility of interlace artifacts can be reduced by anincrease in the picture refresh rate. However, increasing therefresh rate comes at the cost of lowering the vertical resolu-tion. As such, an compromise is created by the 75i format.

Finally, the 100i format is an existing format that is freeof visible interlace artifacts and large area flicker. The dis-advantage is that the vertical resolution has been reduced tothe 625 lines again.

2.2 Test set

Critical image material, originating from either televisioncameras, or obtained from Internet pages has been selectedfor this subjective assessment. In total 4 different still pic-tures were selected, i.e. Grapes, Text, Siena and Web, asshown in Figure 2.

The picturesSiena andGrapes are considered as typicalgood quality scenes from a television camera, whereas thepicturesText andWeb are more typical in a PC like envi-ronment or the Internet. Although all scenes contain spatialhigh frequencies, particularlyText andWeb uses the fre-quency spectrum up to the Nyquist frequency.

As we want to compare the various video formats with-out having to rely on scan-rate conversion quality, we usedstill picture in these experiments only, i.e. we considered anideal scan-rate conversion.

2.3 Test conditions

The characteristics of the display device often limits thescanning or display format. Moreover, various display typeslike Cathode Ray Tube (CRT), Plasma Display Panel (PDP),Liquid Crystal Display (LCD), etc, differ in characteristicsand do, therefore, not necessarily share the same ’optimal’display format. In our experiments we focused on optimiz-ing the display format for the CRT.

Two 21” Philips computer monitor CRTs were used dur-ing both sessions. Because of the high vertical resolutionthat is required to evaluate some of the video formats, com-puter monitor tubes instead of television tubes had to be

(a)Siena

(b)Grapes

(c)Text

(d)Web

Figure 2: Snapshots of the test set.

chosen. This, however, has the disadvantage that the over-all luminance of the display is limited, and therefore, theviewing distance had to be fixed to twice (because of theemulated resolution (see Section 2.4)) the standard viewingdistance for computer monitor applications, which is 0.5 -0.7 m. Thus, subjects assessed all pairs of images at twoviewing distances: 1 m and 1.5 m. In the results we did notfind a difference between these viewing distances. There-fore, we combined the scores of both.

Both CRTs were adjusted to a luminance of 75 cd/m2 ata maximal contrast and to a colour temperature of about7000K (7250K for one monitor and 6700 K for the other).In order to prevent that small differences between both mon-itors affect the results, each pair of images is shown twice,i.e. once with the first image of the pair on the left moni-tor and the second one on the right monitor, and once viceversa. The environmental illumination was 50 lux.

Nineteen (European) subjects ranging in age between 25and 45 years participated in the experiment. Only part ofthe subjects are experienced in video processing, whereasthe remainder can be considered as non-experts. All havea (corrected-to-normal) visus of at least 1 on the LandoltC-scale.

2.4 Optimizing the spot size

A complication, when comparing different scanning for-mats on a single CRT, is that the spot dimensions cannotbe simultaneously optimal for all formats. Clearly a finespot is required to exploit the highest vertical resolution re-sulting from the scanning format with the highest number oflines. However, for the scanning formats with a lower linecount, such a fine spot may lead to an annoying visibility ofthe line structure. To prevent the choice of the spot dimen-sions leading to a bias in the optimization of the scanningformat, we optimized the spot size per scanning format asa balance between perceived sharpness and the visibility ofline structure in a first session of the subjective test.

To realize a variable spot size without changing the dis-play, we emulated a relatively low resolution display on ahigh resolution monitor. A single line was mapped to anumber of scanning lines of the monitor, using a Gaussianfilter to represent the spot dimensions of the emulated CRT.As such, we scaled the video formats mentioned before witha factor of two in both dimensions, and increased the view-ing distance accordingly. For example, the 50p format wasemulated by displaying a ’288 lines by 360 pixels’ pictureon a 576 lines by 720 pixel grid. The additional ’head-room’allowed us to emulate a gaussian spot.

To arrive at the displayed format, we started by properlydownsampling, using a 11 tap FIR filter, the progressivelyscanned pictures and applied another FIR filter in the lineardomain3 to simulate the ideal gaussian spot.

3Filtering in the ’gamma-domain’ would give rise to a difference in

0.50 0.75 1.00 1.25 1.50 1.750.50 97.2 100.0 86.1 80.6 83.30.75 2.8 58.3 38.9 38.9 38.91.00 0 41.7 36.1 33.3 22.21.25 13.9 61.1 63.9 41.7 19.41.50 19.4 61.1 66.7 58.3 22.21.75 16.7 61.1 77.8 80.6 77.8

(a)0.75 1.00 1.25 1.50 1.75 2.00

0.75 80.6 63.9 63.9 55.6 44.41.00 19.4 63.9 36.1 41.7 41.71.25 36.1 36.1 44.4 41.7 38.91.50 36.1 63.9 55.6 52.8 47.21.75 44.4 58.3 58.3 47.2 55.62.00 55.6 58.3 61.1 52.8 44.4

(b)

Table 1: Results of the subjective assessment in % for the a)interlaced format, and the b) progressive format with respectto the ’optimal’ spot dimension.

During the first session of the subjective test subjects as-sessed for two of the four pictures (i.e.Siena andText) sixdifferent spot sizes in a paired-comparison test. The optimalspot size was determined for the 100 Hz interlaced as wellas for the 50 Hz progressive video format, while the op-timal dimensions for the remaining formats were deducedfrom these results. We used a simplified and ideal spot forour emulation, defined by:

G(x) = Ce−(σ x)2(1)

whereC is a constant andσ our ’width’ control parameter.For the interlaced video format theσ varied betweenσ =

0.50, 0.75, 1.00, 1.25, 1.50 and 1.75, whereas for the pro-gressive video format theσ varied in the rangeσ = 0.75,1.00, 1.25, 1.50, 1.75 and 2.00. Note thatσ = 0 is notequal to a spot of zero width, but reflects the intrinsic spotshape of the CRT.

The results of the first session were used to select the bestspot size for each of the five video formats described above.During the second session of the subjective test, the vari-ous video formats with an ’optimized’ spot dimension performat, are evaluated for all four pictures again in a paired-comparison set-up.

3 RESULTS

The results of the first session measuring the subjectivelybest spot dimension are show in Table 1. The percentagesas indicated in the table show the preference of the spot di-mension as indicated at the column head over the spot di-mension indicated at the row head, e.g. there is a 86.1%preference of spot dimension ’defined’ byσ = 1.25 overthe one ’defined’ byσ = 0.50. A shaded table cell indicates

light output for the various filters.

0 0.5 1 1.5 2

0.50

0.75

1.001.25

1.50

1.75

Quality Scale

(a)

−0.4 −0.2 0 0.2 0.4 0.6 0.8

0.75

1.00

1.251.50

1.75

2.00

Quality Scale

(b)

Figure 3: Results of the spot ’optimization’ for the a) inter-laced format, and b) progressive format.

that the corresponding preference is significantly differentfrom 50%, which would be the result when subjects wererandomly selecting just one of the spot dimensions withouthaving a real preference.

These tables of percentages can be transformed towardsz-values using Thurstone’s law [4]. The resulting z-valuesare then summed over the columns, resulting in a qualityscale value, as shown in Figure 3.

The relative distance on the linear quality axis betweenthe various spot sizes as shown in Figure 3 relate to theirperceived differences. As the scale is linear, a distance offor example 2 on the quality axis is perceived as twice asstrong as a difference of 1 on the same axis.

The results for the interlaced format reveals that aσ of1.00 is perceived as optimal for the preselected list, whereasfor the progressive format a slight advantage was found forσ equal to 1.25 over 1.00. As such, a somewhat larger spotseems to be preferred for the interlaced format over the pro-gressive one. The larger spot compensates for the increasedline crawl for the 100i format and the non perfect interlacing

formats 50p 60p 50i 75i 100i50p 40.8 83.5 97.4 84.260p 59.2 84.2 96.7 89.550i 16.5 15.8 71.1 44.775i 2.6 3.3 28.9 21.0

100i 15.8 10.5 55.3 79.0

Table 2: Overall results of the subjective assessment in %.

0 0.5 1 1.5 2

50i

50p

75i

100i

60p

Quality Scale

Figure 4: Overall results of the subjective assessment.

at 100Hz due to the 50Hz magnetic scatter field.The ’optimal’ spot sizes are used in the second session

of the experiment eliminating a bias for either of the videoformats due to a spot profile that is not equally balanced forthe individual formats.

The results of this second session view the preferences forthe individual video formats, as shown in Table 2. Again,we can use Thurstone’s law [4] to compute the z-values, andsummed over the columns results in a quality scale value foreach video format. The resulting quality scale averaged overall four pictures is given in Figure 4. It shows that the 75iformat is the best, followed by the 50i and the 100i formats.The difference in quality between the latter two formats isnot statistically significant. The 50p and 60p formats bothhave the lowest quality, and again, their mutual difference isnot statistically significant. The results can be summarizedas: 75i> (50i > 100i)> (50p> 60p) where the formats be-tween the same parenthesis belong to the same quality level.There were some small differences in the quality scale val-ues for a given video format amongst the four pictures. Theresults demonstrated that the preference for the 75i formatwas most noticed for the pictures that contain highly tex-tured regions (mainly found in graphical or Internet relatedpictures), which was expected as interlace artifacts start tobecome most pronounced in highly detailed pictures. Nev-ertheless, it resulted in the same grouping and ranking of thevideo formats for each of the four pictures individually.

From the subjective evaluation, we draw the conclusionthat the video format of 75i is superior to the alternativescannings. A remarkable observation is that the preference

is most impressive in comparing the 75i format with thetwo progressive display formats (50p and 60p). In fact,all the interlaced formats are preferred over the progressiveformats. The loss of vertical resolution of the progressiveformats is apparently recognized as the most distinguishingelement.

Another interesting observation is that the preference ofthe 75i format over the 100i format indicates that the ob-servers clearly notice the difference in vertical resolution,and probably hardly observe any difference in large areaflicker or line flicker. The difference of the 100i with the50i format was not considered to be significant, i.e. it wasfound difficult to chose between a picture with significantline / large area flicker and a high vertical resolution, anda picture with a significant lower resolution and no visibleinterlace artifacts.

4 CONCLUSIONS

Recent progress in scan-rate conversion technology enablesa decoupling of the received video format and the displayformat. As a result, we can choose the displayed number ofscanning lines, the interlace factor and picture rate at will.

Due to this new freedom, there is a need to investigatethe ’optimal’ display format for a given application. In thispaper, we evaluated various display formats for a CRT dis-play.

As the spot dimension can not be simultaneously opti-mal for all the display formats, we started with subjectivelyinvestigating the best dimension for both an interlaced anda progressive format. The results of this experiment wereused in the subjective assessment to determine the ’optimal’display format for a CRT display.

Our subjective evaluation revealed that our viewers al-ways preferred interlaced scanning over progressive scan-ning with the same pixel rate. From the range of picturerates that we tested, the 75i format turned out to providethe best balance between flicker and resolution. Finally, ourviewers preferred the 100i high-end television format overthe 60p high-end format. Of all tested formats, we concludethat the interlaced format at 75 Hz, 833 lines is optimal forCRT displays with a 27 MHz luminance pixel rate.

REFERENCES

[1] G. de Haan, J. Kettenis and B.D. Loore, ‘IC for motion-compensated 100 Hz TV with natural-motion movie-mode ’,IEEE Tr. on Consumer Electronics, May 1996,pp. 165-174.

[2] G. de Haan, ‘IC for Motion-Compensated De-interlacing, Noise Reduction, and Picture-Rate Conver-sion’, IEEE Tr. on Consumer Electronics, Vol. 45, Au-gust 1999, pp. 617-624.

[3] M. Schu, G. Scheffler, C. Tuschen, and A. Stolze, ‘Sys-tem on silicon-IC for motion compensated scan rateconversion, picture-in-picture processing, split screenapplications and display processing’,IEEE Tr. on Con-sumer Electronics, Vol. 45, August 1999, pp. 842-850.

[4] P.G. Engeldrum, ’Psychometric scaling: a toolkit forimaging systems development’, (Imcotek Press, Winch-ester, 2000).

Erwin Bellers was born in Enschede,The Netherlands, on November 14,1965. He received his B.Sc degree in1987 from the‘Technische Hogeschool’,his M.Sc degree with distinction fromthe University of Twente in 1993,and his Ph.D degree from theDelftUniversity of Technologyin 2000 with

his thesis about de–interlacing. In 1993, Erwin joinedPhilips Research Laboratoriesin Eindhoven as a ResearchScientist. In 2000, Erwin joined Philips Research USAas a Senior Scientist. His major interest is in video en-hancement algorithms, including resolution enhancement,motion–compensated de–interlacing, motion estimation,and motion–compensated picture–rate up–conversion. Hereceived the second price in the 1997 ICCE OutstandingPaper Awards program, and was invited as a co–authorfor the Proceedings of the IEEE. His work has resulted inabout 15 patents and patent applications, about 20 papers,and 2 books so far. Erwin is member of the IEEE.

Ingrid Heynderickx was born inSint-Niklaas, Belgium, on the 28thof October 1961. She received fromthe University of Antwerp her M.Scdegree in Physics in 1983 and her PhDdegree in Physics in 1986. In 1987, shejoined Philips as a research scientist,and meanwhile worked in different areas

of research: optical characteristics of displays, processingof liquid crystalline polymers and the evaluation of thefunctionality of personal care devices. Since 1999, she isa principal scientist in the Video Processing and VisualPerception group at Philips Research Eindhoven, whereshe is heading the project on Visual Perception of DisplaySystems.

Gerard de Haan received the B.Sc.,the M.Sc., and the Ph.D. degree fromDelft University of Technology in 1977,1979 and 1992 respectively. He joinedPhilips Research in 1979. Currentlyhe is a Research Fellow in the groupVideo Processing & Visual Perception ofPhilips Research Eindhoven and a Pro-fessor at the Eindhoven University of

Technology. He has a particular interest in algorithmsfor motion estimation, scan rate conversion, and imageenhancement. His work in these areas has resulted inseveral books, about 70 papers (www.ics.ele.tue.nl/ de-haan/publications.html), some 50 patents and patent appli-cations, and several commercially available ICs. He wasthe first place winner in the 1995 ICCE Outstanding PaperAwards program, the second place winner in 1997 and in1998, and the 1998 recipient of the Gilles Holst Award. ThePhilips ‘Natural Motion Television’ concept, based on hisPhD-study received the European Innovation Award of theYear 95/96 from the European Imaging and Sound Associ-ation. Gerard de Haan is a Senior Member of the IEEE.

Inge de Weerd received the B.Sc. de-gree in Engineering Physics from the Ri-jswijk Institute of Technology (Technis-che Hogeschool Rijswijk), the Nether-lands, in 1996. In 1999, she receivedher MSc degree in Technology & Soci-ety (focus on Human-Technology Inter-action) from the Technical University of

Eindhoven. She joined Philips Research Laboratories inEindhoven in April 2000 as a research scientist in the VideoProcessing and Visual Perception Group. The emphasis ofher work is on perceptual evaluation of image quality.