Video decoding system and method having post-processing to reduce sharpness prediction drift

1. A video processing system comprising:

a decoder having a complexity reduction system; and a post-processing system for processing an output of the decoder, wherein the post-processing system includes a filter for intentionally blurring a set of frames in a group of pictures (GOP) to achieve a smooth visual transition between frames, wherein the post-processing system further includes a controller for maintaining a complexity of the post-processing system within a resource budget and for calculating a cut-off frequency of the filter; and a frame selector for selecting the set of frames in the GOP to be blurred by the filter,

wherein the controller calculates a number of frames N it is capable of processing for a given resource constraint r according to the equation:

N=r/(p·f·k), wherein p is a picture size, f is the filter length, and k is a constant representing a computational resource needed per pixel per filter coefficient.

2. The video processing system of claim 1 , wherein the set of frames selected by the frame selector to be blurred comprises the first N frames in the GOP.

3. A video processing system comprising:

a decoder having a complexity reduction system; and a post-processing system for processing an output of the decoder, wherein the post-processing system includes a filter for intentionally blurring a set of frames in a group of pictures (GOP) to achieve a smooth visual transition between frames, wherein the post-processing system further includes a controller for maintaining a complexity of the post-processing system within a resource budget and for calculating a cut-off frequency of the filter; and a frame selector for selecting the set of frames in the GOP to be blurred by the filter,

wherein the controller estimates the cut-off frequency C using the equation C=C 0 −(S·N r ), wherein C 0 is a constant, S measures the sharpness of a first I frame in the GOP, and N r is the number of reference frames in the GOP.

4. A method for eliminating prediction drift in a complexity reduced video sequence, the method comprising:

determining an amount of post-processing resources available;

determining a number of frames in the video sequence that can be processed pursuant to the available resources;

selecting a set of frames within the video sequence to process;

filtering the set of selected frame, wherein the filtering blurs the set of frames in order to achieve a smooth visual transition between frames; and

calculating a cut-off frequency of the filter for the video sequence prior to the filtering operation,

wherein the cut-off frequency C is calculated using the equation: C=C 0 −(S·N r ), wherein C 0 is a constant, S measures the sharpness of a first I frame in the GOP, and N r is the number of reference frames in the GOP.

5. The method of claim 4 , wherein the sharpness S is calculated based on the arrangement of non-zero discrete cosine transform (DCT) coefficients in each of a plurality of DCT blocks of the first I frame in the GOP.

6. A method for eliminating prediction drift in a complexity reduced video sequence, the method comprising:

determining an amount of post-processing resources available;

determining a number of frames in the video sequence that can be processed pursuant to the available resources;

selecting a set of frames within the video sequence to process; and filtering the set of selected frame, wherein the filtering blurs the set of frames in order to achieve a smooth visual transition between frames, wherein the number of video frames N that can be processed under a given resource constraint r is calculated according to the equation:

N=r/(p·f·k), wherein p is a picture size, f is a filter length, and k is a constant representing a computational resource needed per pixel per filter coefficient.

7. The method of claim 6 , wherein the set of frames selected to be blurred comprises the first N frames in the video sequence.

8. A post-processing system for eliminating prediction drift in a complexity reduced video sequence, comprising:

a controller for receiving an available amount of post-processing resources and determining a number of frames in the video sequence that can be processed pursuant to the available resources;

a frame selector for selecting a set of frames within the video sequence to process; and

a filter for filtering the set of frames, wherein the filtering blurs the set of frames in order to achieve a smooth visual transition between frames, wherein the controller calculates a cut-off frequency of the filter for the video sequence,

wherein the cut-off frequency C is calculated using the equation: C=C 0 −(S·N r ), wherein C 0 is a constant, S measures the sharpness of a first I frame in the GOP, and N r is the number of reference frames in the GOP.

9. The post-processing system of claim 8 , wherein the sharpness S is calculated based on the arrangement of non-zero discrete cosine transform (DCT) coefficients in each of a plurality of DCT blocks of the first I frame in the GOP.

10. A post-processing system for eliminating prediction drift in a complexity reduced video sequence, comprising:

a controller for receiving an available amount of post-processing resources and determining a number of frames in the video sequence that can be processed pursuant to the available resources;

a frame selector for selecting a set of frames within the video sequence to process; and

a filter for filtering the set of frames, wherein the filtering blurs the set of frames in order to achieve a smooth visual transition between frames, wherein controller calculates the number of video frames N that can be processed under a given resource constraint r according to the equation: N=r/(p·f·k), wherein p is a picture size, f is a filter length, and k is a constant representing a computational resource needed per pixel per filter coefficient.

11. The post-processing system of claim 10 , wherein set of frames selected to be blurred by the frame selector comprises the first N frames in the video sequence.