Television – Monitoring – testing – or measuring – Transmission path testing
Reexamination Certificate
1997-01-02
2001-05-29
Faile, Andrew (Department: 2611)
Television
Monitoring, testing, or measuring
Transmission path testing
C348S192000, C348S180000
Reexamination Certificate
active
06239834
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to an apparatus for evaluating digital picture quality of a coding device using digital picture compression techniques, particularly to an apparatus for evaluating digital picture quality of a digital video codec (coder-decoder) or a digital image transmission path.
DESCRIPTION OF THE RELATED ART
The assessment of such picture quality can be classified to two methods of subjective evaluation and objective evaluation.
As for the subjective evaluation, there is for example a method of evaluating television picture quality in accordance with ITU-R BT.500-7 Recommendations. According to this method, the television picture is subjectively evaluated by at least 15 assessors for several days. Thus, this subjective evaluation method has problems that work-load accompanied with the test session will become extremely heavy and therefore the cost necessary for the assessment will become very high even when only a few pictures are evaluated. In addition, this subjective evaluation method needs a very long period of time for obtaining the result due to its heavy work-load. Therefore, no assessment of picture quality is possible within a short time period.
As for a conventional objective evaluation, there is a typical method using a SNR (Signal to Noise Ratio) which is calculated from simple differences, such as mean square errors, between pixel signals in an original picture and corresponding pixel signals in a picture to be evaluated. This objective evaluation method can derive evaluated result within a short time period. However, because no human visual perception characteristics is considered, the derived result of evaluation according to this method may be frequently differed from the result of the subjective evaluation method.
Another objective evaluation method using a WSNR (Weighted Signal to Noise Ratio) which is weighted in accordance with a human visual perception, considering that the difference between the results of the evaluation using the SNR and of the subjective evaluation is caused by visually conspicuous noises generated on the background, is known. The objective evaluation methods using the WSNR are described for example in S. Hangai, T. Wada and K. Miyauchi, “Experimentally Derived Weighting Function in Spatial Frequency Domain for Evaluating Still Picture”, Proc. of The Institute of Image Information and Television Engineers, Vol. 46, No. 3, pp. 295-299, 1992 and T. Hamada, S. Matsumoto and in H. Murakami, “The Marginal Performance of Motion Compensated DCT Coding Scheme Based on Optimized Quantizer”, Proc. of The Institute of Electronics, Information and Communication Engineers, B-I, Vol.J78-B-I, No. 11, pp. 689-699, Jan. 1995.
According to this known objective evaluation method, it is considered that whether generated noises are visually conspicuous with respect to the background of each block in the frame. However, since no visual perception characteristics with respect to the background of the whole frame is considered in this method, the evaluated result represented by the obtained WSNR may greatly differ from that of the subjective evaluation.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide a low cost apparatus for evaluating digital picture quality, whereby evaluation can be achieved in a short period of time and also the evaluated result is close to that of the subjective evaluation.
According to the present invention, an apparatus for evaluating digital picture quality includes a SNR calculation module for calculating a SNR based upon differences between data of video picture to be evaluated and data of original video picture, a module for calculating a power of AC component of overall data of one of the video pictures in a frame, and a weighting module for weighting the calculated SNR based upon the calculated power of the AC component.
Since the evaluation apparatus according to the present invention considers for evaluation of picture quality the characteristics of visual perception with respect to the background over the whole picture by weighting the SNR with the power of AC component of overall video data of the picture, it is possible to finally obtain an excellent evaluated result which is closer to that of the subjective evaluation. If evaluating picture quality base upon a SNR (WSNR) considering only visual perception characteristics for each block in a frame, its result will differ from the actual picture quality, depending upon whether that block is noticeable over the whole frame or not. Thus, by introducing the power of AC component of overall video data of the picture, which power represents the global activity of the whole frame, the amount of visual striking with respect to each block can be taken into account causing the evaluated result to close to the subjective evaluation result.
It is preferred that the SNR calculation module includes a module for calculating a SNR overall the frame by calculating, for each block of the frame, differences between the data of the video picture to be evaluated and the data of the original video picture.
It is also preferred that the apparatus further includes a coefficient calculation module for calculating weighting coefficients used for weighting the data of the video picture to be evaluated and the data of the original video picture for each block in the frame based upon characteristics of human visual perception, and that wherein the SNR calculation module calculates the SNR based upon the differences by weighting with the weighting coefficients from the coefficient calculation module.
The coefficient calculation module may include a module for calculating the weighting coefficients for each block in the frame based upon the characteristics of visual sensitivity across spatial frequencies. Since high-frequency noise is less perceptible, the evaluated result obtained by weighting based upon the characteristics of visual sensitivity across spatial frequencies will be close to that of the subjective evaluation.
The coefficient calculation module may include a module for calculating the weighting coefficients for each block in the frame based upon the noise masking effect. Since noise in high activity block is less perceptible, the evaluated result obtained by weighting based upon the noise masking effect will be close to that of the subjective evaluation.
It is preferred that the weighting module includes a module for weighting using a mapping function with coefficients each of which is predetermined based upon the calculated power of the AC component.
According to the present invention, furthermore, an apparatus for evaluating digital picture quality includes a module for providing gaze point information which indicates a position of a gaze point related to a video picture to be evaluated, and a SNR calculation module for calculating a SNR of pixels at and around the gaze point position indicated by the gaze point information, based upon differences between data of the video picture to be evaluated and data of a original video picture.
Since the evaluation apparatus obtains the evaluated result of picture quality by calculating a SNR based upon differences between video data values and original video data values with respect to pixels located at and around the gaze point. Thus, it is considered whether noise occurs at a relatively striking position or nonstriking position in the picture causing the evaluated result to close to that of the subjective evaluation.
It is preferred that the apparatus further includes a module for weighting the data of the video picture to be evaluated and the data of the original video picture with respect to pixels at and around the gaze point position indicated by the provided gaze point information based upon the characteristics of visual sensitivity around the centered gaze point, and that the SNR calculation module calculates the SNR based upon the differences between the weighted data of the video picture to be evaluated and the weighted data of the original video picture. Thus,
Hamada Takahiro
Matsumoto Shuichi
Miyaji Satoshi
Arent Fox Kintner & Plotkin & Kahn, PLLC
Faile Andrew
Kokusai Denshin Denwa Co. Ltd.
Srivastava Vivek
LandOfFree
Apparatus for evaluating digital picture quality does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus for evaluating digital picture quality, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus for evaluating digital picture quality will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2479302