Method and apparatus for applying accurate parameter estimation and efficient fade detection in weighted prediction for video compression
View Patent ↗A method to control weighted prediction for video compression. The method comprises (A) generating statistics based upon analysis of a plurality of video frames, (B) detecting a fade condition based upon the statistics, (C) generating one or more estimated weighted prediction parameters based upon the statistics when the fade condition is detected and (D) encoding the plurality of video frames. The encoding takes into account the estimated weighted prediction parameters when generated.
1. A method to control weighted prediction for video compression comprising:
generating statistics based upon analysis of at least one of a plurality of characteristics of a plurality of video frames, the plurality of characteristics comprising luminance, chrominance, spatial and temporal characteristics of the plurality of video frames;
detecting a fade condition based upon said statistics that are associated with the at least one of the plurality of characteristics, the fade condition being based on a regular change of said statistics among the plurality of video frames;
generating one or more estimated weighted prediction parameters based upon said statistics when said fade condition is detected;
encoding said plurality of video frames, wherein said encoding takes into account said estimated weighted prediction parameters when generated; and
wherein the statistics associated with the at least one of a plurality of characteristics comprises an average value per picture, a variance value per picture, a derivative of the average value per picture, and a ratio of the variance value per picture for a current picture and each associated reference picture.
2. The method according to claim 1 , wherein said one or more estimated weighted prediction parameters comprise a weighting factor and an additive offset.
3. The method according to claim 1 , wherein detecting said fade condition comprises:
detecting a change in luminance level;
detecting a steady change between pictures;
detecting a consistent change among said pictures; and
detecting changes in a ratio of variance.
4. The method according to claim 1 , wherein generating said one or more estimated weighted prediction parameters comprises:
calculating weights for each picture in a prediction list;
clipping the weights; and
calculating offset values.
5. The method according to claim 4 , wherein said weights are calculated from a square root of a ratio of variances of a current picture and a corresponding reference picture.
6. The method according to claim 4 , wherein said offset values are calculated based upon said weights and average luminance statistics of a current picture and a corresponding reference picture.
7. The method according to claim 4 , wherein clipping the weights comprises:
comparing each of said weights with a first predetermined threshold value and a second predetermined threshold value;
setting each of said weights that is less than said first predetermined threshold value to said first predetermined threshold value; and
setting each of said weights that is greater than said second predetermined threshold value to said second predetermined threshold value.
8. The method according to claim 1 , wherein the statistics generated comprise an average luminance value per picture and a luminance variance value per picture.
9. The method according to claim 8 , further comprising generating a first derivative of the average luminance value per picture statistic, a second derivative of the average luminance value per picture statistic and a ratio of the luminance variance value per picture statistic for a current picture and each associated reference picture.
10. An apparatus comprising:
means for generating statistics based upon analysis of at least one of a plurality of characteristics of a plurality of video frames, the plurality of characteristics comprising luminance, chrominance, spatial and temporal characteristics of the plurality of video frames;
means for detecting a fade condition based upon said statistics that are associated with the at least one of the plurality of characteristics, the fade condition being based on a regular change of said statistics among the plurality of video frames;
means for generating one or more estimated weighted prediction parameters based upon said statistics when said fade condition is detected; and
means for encoding said plurality of video frames, wherein said encoding takes into account said estimated weighted prediction parameters when generated; and
wherein the statistics associated with the at least one of a plurality of characteristics comprises an average value per picture, a variance value per picture, a derivative of the average value per picture, and a ratio of the variance value per picture for a current picture and each associated reference picture.
11. An apparatus comprising:
a video analyzer circuit configured to generate one or more statistics based upon an analysis of at least one of a plurality of characteristics of a plurality of video frames, the plurality of characteristics comprising luminance, chrominance, spatial and temporal characteristics of the plurality of video frames;
a detector circuit configured to detect a fade condition based upon said one or more statistics that are associated with the at least one of the plurality of characteristics, the fade condition being based on a regular change of said statistics among the plurality of video frames;
an estimator circuit configured to generate one or more weighted prediction parameters in response to said one or more statistics and a state of said plurality of video frames;
an encoder configured to encode said plurality of video frames using said one or more weighted prediction parameters when generated; and
wherein the statistics associated with the at least one of a plurality of characteristics comprises an average value per picture, a variance value per picture, a derivative of the average value per picture, and a ratio of the variance value per picture for a current picture and each associated reference picture.
12. The apparatus according to claim 11 , wherein said one or more weighted prediction parameters comprise a weighting factor and an additive offset.
13. The apparatus according to claim 11 , wherein said detector circuit is further configured to:
detect a change in luminance level;
detect a steady change between pictures;
detect a consistent change among said pictures; and
detect changes in a ratio of variance.
14. The apparatus according to claim 11 , wherein said detector circuit is further configured to:
set a state of a current video frame to a first value in response to detecting all of a change in luminance level, a steady change between pictures, a consistent change among said pictures and changes in a ratio of variance; and
set said state of said current video frame to a second value in response to a lack of detection of any one of a change in luminance level, a steady change between pictures, a consistent change among said pictures and changes in a ratio of variance.
15. The apparatus according to claim 11 , wherein generating said one or more weighted prediction parameters comprises:
calculating weights for each picture in a prediction list;
clipping the weights; and
calculating offset values.
16. The apparatus according to claim 15 , wherein said weights are calculated from a square root of a ratio of variances of a current picture and a corresponding reference picture.
17. The apparatus according to claim 15 , wherein clipping the weights comprises:
comparing each of said weights with a first predetermined threshold value and a second predetermined threshold value;
setting each of said weights that is less than the first predetermined threshold value to said first predetermined threshold value; and
setting each of said weights that is greater than said second predetermined threshold value to said second predetermined threshold value.
18. The apparatus according to claim 15 , wherein said offset values are calculated based upon said weights and average luminance statistics of a current picture and a corresponding reference picture.
19. The apparatus according to claim 11 , wherein the statistics comprise an average luminance value per picture and a luminance variance value per picture.
20. The apparatus according to claim 19 , wherein the video analyzer circuit is further configured to generate a first derivative of the average luminance value per picture statistic, a second derivative of the average luminance value per picture statistic and a ratio of the luminance variance value per picture statistic for a current picture and each associated reference picture.