NHK Laboratories Note No. 457

EVALUATION OF MOTION-PICTURE QUALITY ON 42-IN.PDP

by
Toshimitsu Koura, Toshihiro Yamamoto, Keiji Ishii,
Yoshimichi Takano, Masahiko Seki, Hideki Kokubun*, Taiichiro Kurita**,
Hiroshi Murakami* * *, Kiichi Kobayashi* * *, and Koichi Yamaguchi***
(Display and Optical Devices Research Division)
*Advanced Imaging Devices Research Division
* *Three Dimensional Audio-visual Systems Research Division
* * *NHK Engineering Service

ABSTRACT
The effect of new signal processing and driving methods introduced to a 42-in. HDTV DC PDP to improve motion-picture quality is verified by both evaluation by computer simulation and subjective evaluation of the motion pictures. The results show that our techniques sufficiently improve picture quality for practical use in the HDTV PDP receiver.
1. INTRODUCTION
The HDTV has higher resolution than the current broadcasting systems and it is possible to enjoy marvelous force and presence when viewed on a large screen of over 40- in. The plasma display panel (PDP) is a powerful candidate for use in the large wall hanging HDTV, and recently a display of over 40-in. has been developed[1][2]. In order to spread the use of HDTV receivers in ordinary homes, we have especially been conducting research to realize the wall hanging HDTV. We developed a 42-in. diagonal, high resolution DC PDP as the first step toward developing a practical HDTV PDP in 1996[3].
One serious problem conceming PDP picture quality is the degradation of motion pictures caused by the method used to represent gray levels. In particular, since the HDTV system requires high-quality pictures, this degradation must be reduced to an almost imperceptible level for the practical HDTV PDP. This degradation, called false contour, is caused by both the weighted sub-field emission scheme and an effect produced by following the motion pictures with the eyes[4]. Although several kinds of signal processing technique have been reported for reducing the degradation of motion pictures[1][5], such reports were not about HDTV pictures and the effects of these techniques have not yet been evaluated quantitatively.
We have already developed new signal processing and driving methods for DC PDP which reduce this degradation[6]. We have also been able to confirm the improvement of motion-picture quality using these methods, by displaying many motion pictures on a panel, but had not evaluated the picture quality quantitatively until now.
This paper discusses our quantitative verification of motion-picture quality by both S/N calculation with computer simulation and a subjective evaluation using a 42-in.DCPDP.

2. DEGRADATION OF THE MOTION PICTURE
Generally, the PDP represents gray scale with sub-fields. This method places the light emission pulse with the period weighted by a power of two in 1 field and obtains the gray level according to the integral visual effect. Fig.1 shows the conventional emission pattern on a DC PDP.


Fig.1 Conventional emission pattern

With this emission pattern, assume that a picture of 128 picture level on the left and 127 picture level on the right moves to the right. Fig.2 shows the emission pattern on the panel. The horizontal axis of Fig.2 shows the horizontal position of the pixel, and the vertical axis shows the time. At this time, 1uminance is perceived to be higher than the correct value because of integration along the continuous line, as shown in Fig.2, by the following eyes.


Fig.2 Example of emission pattern on the panel

Fig.3 shows the results of a picture simulation for the conventional emission scheme (Fig.1), where one still picture is in motion to the right at a constant speed on the PDP. A lot of false contour noise is perceived in the figure. This is very noticeable in pictures with gentle luminance slopes, such as those of human faces and blue skies.

Fig.3 Photograph of false contour simulation

3. IMPROVEMENT TECHNIQUES
We adopted the following teclmiques to improve motion-picture quality on the 42-in. DCPDP.
3.1 TIME COMPRESSION OF THE LIGHT EMISSION PERIOD OF SUB-FIELDS IN A FIELD[6]
The false contour of motion pictures increases as the light emission of each sub- field is widely dispersed in a field. The conventional pattern uses 8 sub-fields, which are dispersed in the field as shown in Fig.1, resulting in the perception of much false contour.
Therefore, we have adopted tlme compression of the light emission period. The light emission pattern in this technique, which uses 8 sub-fields, is shown in Fig.4. The total light emission period is compressed to about twenty percent of the field time.

Fig.4 Compression of emission period
3.2 DIVISION OF UPPER TWO SIGNAL BITS INTO FOUR SUB-FIELDS[6]
The false contour tends to appear when a heavily weighted bit is changed. We divided the upper 2 bits into 4 sub-fields, each of which has a lower weight of 48. This techmique originally required 10 sub-fields to display 256 gray levels, but driving became difficult because the larger number of sub-fields necessitated high speed access. We were able to reduce the number of sub-fields to 9 by adopting error diffusion processing. The emission pattern for this technique is shown in Fig.5.

Fig.5 Compression of emission period and division of upper bits
3.3 ADOPTION OF SUB-FIELD CONTROL(SFC)[6]
A delicate luminance difference appears due to the difference of the upper sub- field pattern even with equal picture levels of neighboring cells. This is perceived as degradation including the false contour of the picture quality. The emission patterns of the sub-flelds are shown in Table 1 for the respective picture levels. In this scheme, there are two possibilities for reproducing the 48 picture level, namely, emission at only the 48 picture level or emission at two sub-fields, the 32 and 16 picture levels. The emission pattern for either mode 1 or 2 is determined by the table, so that two neighboring cells can emit the same pattern as far as possible.

Table 1 Emission pattern for sub-fleld

4. EVALUATION BY COMPUTER SIMULATION
4.1 EVALUATION METHOD
The quantity of false contours is decided by the light emission pattern and the motion speed of the eyes following the object. We make a picture in which a still picture moves horizontally and at a constant velocity, and perform the quantitative evaluation of motion-picture quality by an SM ratio calculation.
The following are the calculation methods.
(1) It is assumed that the still picture moves in the right direction at a constant velocity, and the light emission is integrated along the locus of the following eyes, as shown in Fig.6. It is calculated at the each pixel.
(2) The noise is the level difference of each pixel between the picture obtained by calculation and the original picture.
(3) The noise is obtained by calculating the square mean value of each noise item in the position.
(4) The S/N ratio is obtained from calculations (1) and (3).

Fig.6 Calculation method

Table2 Emission conditions for evaluation
ConditionNumber of
Sub-field
Compressed light
emission period
SFCEmission pattern
A8NoNoFig.1
B8YesNoFig.4
C9YesNoFig.5
D9YesYesFig.5

4.2 SIMULATION RESULTS
We evaluated the four emission conditions shown in Table 2. We used the ITE standard charts, Woman with Carnations and Eiffe1, as the pictures for evaluation.
The simulation results for the 2 pictures are shown in Figs.7(a) and (b). The horizontal axis of Fig.7 shows the motion speed of the picture, and the vertical axis shows the S/N ratio.
The S/N ratios deteriorate as motion speed increases in either condition.
The S/N ratios have been improved remarkably at the emission conditions B, C, D using the compressed emission period.
The 9 sub-field system improves the SM ratios more than the 8 sub-field system, furthermore, adopting SFC improves picture quality.


Fig.7 Simulation results

5. SUBJECTIVE EVALUATION
5.1 EVALUATION METHOD
We also performed subjective evaluations using the double-stimulus impairment scale method ( the European Broadcasting Union (EBU) method ). Table 3 shows the evaluation conditions. We also used the ITE standard charts for the evaluation experiment. We evaluated the four emission conditions shown in Table 2 and motion speed using each of the eight patterns.

Table 3 Evaluation conditions
Evaluation methodThe EBU method
Evaluation scaleFive-grade impairment scale
Display system42-inch DC PDP
SourceHDTV 1-inch D-VTR
Ratio of viewing distance
to picture height
3H
Peak luminance150cd/m2
Arrangement of observersWithin 30 horizonta11y
Other
observation conditions
Conform to standard
observation conditions
Observers15 people (non-expert)


5.2 EVALUATION RESULTS
Figs.8(a) and (b) show the results of subjective evaluation. The horizontal axis of Fig.8 shows the motion speed of the picture, and the vertical axis shows the five-grade impairment scale. Positive motion speed in Fig.8 is where the picture is moved to the right and negative motion speed is where the picture is moved to the left.
The evaluation value (the score) deteriorates as the motion speed increases under all emission conditions. The scores become below the detection limit (a score of 4.5) under all emission conditions at a speed of 20 (cells/field). This seems to be due to factors such as afterglow of the phosphor and movement blur.
Emission condition B has improved picture quality over 2 grades higher than for emission condition A, showing that the time compression of the light emission period was effective.
Emission condition C improves the score more than emission condition B.
There is no clear difference between emission conditions C and D. It is necessary to perform evaluation on actual motion pictures to confirm clearly the effects of sub-field control, but emission condition D obtains the best score at alrriost evaluated motion speed. Emission condition D improves the evaluation results to above the acceptable limit ( a score of 3.5 ) for the full range of evaluated motion speed.
These results confirm that the 42-in. DC PDP using emission condition D can represent motion pictures with sufficient picture quality for the HDTV.


Fig.8 Subjective evaluation results

6. CONCLUSION
  To confirm the effects of the improvement techniques for picture quality adopted in the DC PDP, we performed evaluations by computer simulation and subjective evaluation. The results showed the picture quality under the emission scheme with the compressed emission period was much higher than that under the conventional scheme, and the division of upper bits was also effective for improving motion-picture quality. Based on these results, our evaluation of motion-picture quality confirms that the techniques introduced for the 42-in. DC PDP achieve a sufficient improvement in quality for use in HDTV.

REFERENCE

[1] T.Makino, A.Mochizuki, M.Tajima, T.Ueda, K.Ishida, K.Kariya, Improvement of Video Image Quality in AC-Plasma Display Panels by Suppressing the Unfavorable Coloration Effect with Sufficient Gray Shades Capability, Proc. Asia display'95, pp381-384, October (1995).
[2] T.Nakamura, K.Iseki, Y.Sano, K.Nunomura, Drive for 40-in.-Diagonal Full-Color ac Plasma Display, Proc. SID'95, pp.807-810, May (1995).
[3] K.Ishii, M.Seki, T.Kuriyama, T.Kato, T.Takei, T.Kawai, Y.Murakami, S.Ueda, T.Tajima, J.Koike, H.Murakami, Y.Sasaoka, T.Atsumi, H.Sakamoto, U.Miyagawa, K.Kiriyama, S.Uemura, K.Takahashi, K.Wami, High-Resolution 40-inch Color Plasma Display for HDTV, Proc. Euro Display'96, pp.43-46, October (1996).
[4] T.Masuda, T.Yamaguchi, S.Mikoshiba, NEW CATEGORY CONTOUR NOISE OBSERVED IN PULSE-WIDTH-MODULATED MOVING IMAGES, Proc. 14th IDRC, pp.357-360, October (1994).
[5] T.Yamaguchi, K.Toda, S.Mikoshiba, Improvement in PDP Picture Quality by Three-Dimensional Scattering of Dynamic False Contours, Proc. SID'96, pp.291-294, May (1996).
[6] T.Yamamoto, Y.Takano, K.Ishii, T.Koura, H.Kokubun, K.Majima, T.Kurita, K.Yamaguchi, K.Kobayashi, H.Murakami, Improvement of Moving-Picture Quality on 42-in.-Diagonal PDP for HDTV, Proc. SID'97, pp.217-220, May (1997).


Copyright 1999 NHK (Japan Broadcasting Corporation) All rights reserved. Unauthorized copy of the pages is prohibited.

BackHome