Integerization for interpolation filter design in video coding

Patent metadata

Field	Value
Publication number	US-12519937-B2
Application number	US-202318842334-A
Country	US
Kind code	B2
Filing date	Mar 2, 2023
Priority date	Mar 4, 2022
Publication date	Jan 6, 2026
Grant date	Jan 6, 2026

Abstract

In some embodiments, a video encoder encodes a video into a video bitstream. The video encoder accesses a set of frames of the video and performs inter prediction for the set of frames using a set of integerized interpolation filters to generate prediction residuals to be encoded into the video bitstream. The set of integerized interpolation filters are generated by integerizing a set of interpolation filters, each of the set of interpolation filters having floating-point filter coefficients. For each interpolation filter, two integerized filter coefficient values are generated for each filter coefficient and a set of filter candidates are generated based on the two integerized values for each filter coefficient. An error metric for each filter candidate is calculated and an integerized interpolation filter having the lowest error metric is selected for the interpolation filter from the set of filter candidates.

First claim

The invention claimed is:

1 . A method for encoding a video, the method comprising:
accessing a plurality of frames of the video; performing inter prediction for the plurality of frames using a set of integerized interpolation filters to generate prediction residuals for the plurality of frames, wherein the set of integerized interpolation filters is generated by: accessing a set of interpolation filters, each of the set of interpolation filters having floating-point filter coefficients; for each interpolation filter in the set of interpolation filters, generating two integerized filter coefficient values for each filter coefficient of the interpolation filter, generating a set of filter candidates based on the two integerized filter coefficient values for each filter coefficient, calculating an error metric for each filter candidate in the set of filter candidates, and selecting an integerized interpolation filter for the interpolation filter from the set of filter candidates, the selected integerized interpolation filter having a lowest error metric among the set of filter candidates; and encoding the prediction residuals for the plurality of frames into a bitstream representing the video.

2 . The method of claim 1 , wherein the error metric is defined as a squared error of the integerized filter coefficients of the filter candidate from the corresponding floating-point filter coefficients scaled by a particular value.

3 . The method of claim 1 , wherein the error metric is calculated by approximating an integral of a spectral error between a first frequency response of the filter candidate and a second frequency response of an ideal interpolator.

4 . The method of claim 3 , wherein the error metric is further defined by applying weights to the spectral error.

5 . The method of claim 1 , wherein selecting the integerized interpolation filter from the set of filter candidates comprises:
selecting a reduced set of filter candidates from the set of filter candidates, each filter candidate in the reduced set of filter candidates having an error metric lower than remaining filter candidates in the set of filter candidates; and determine a rate-distortion result for each of the reduced set of filter candidates by applying respective filter candidates in a video coding system, wherein the integerized interpolation filter is selected from the reduced set of filter candidates based on the rate-distortion results.

6 . The method of claim 1 , wherein two integerized filter coefficient values for each filter coefficient comprise a largest integer value that is smaller than the filter coefficient scaled by a particular value and a smallest integer value that is larger than the filter coefficient scaled by the particular value.

7 . The method of claim 6 , wherein the set of filter candidates comprise filter candidates having filter coefficients selected from the two integerized filter coefficient values of respective filter coefficients.

8 . A non-transitory computer-readable medium having program code that is stored thereon, the program code executable by one or more processing devices for performing operations comprising:
accessing a plurality of frames of a video; performing inter prediction for the plurality of frames using a set of integerized interpolation filters to generate prediction residuals for the plurality of frames, wherein the set of integerized interpolation filters is generated by: accessing a set of interpolation filters, each of the set of interpolation filters having floating-point filter coefficients; for each interpolation filter in the set of interpolation filters, generating two integerized filter coefficient values for each filter coefficient of the interpolation filter, generating a set of filter candidates based on the two integerized filter coefficient values for each filter coefficient, calculating an error metric for each filter candidate in the set of filter candidates, and selecting an integerized interpolation filter for the interpolation filter from the set of filter candidates, the selected integerized interpolation filter having a lowest error metric among the set of filter candidates; and encoding the prediction residuals for the plurality of frames into a bitstream representing the video.

9 . The non-transitory computer-readable medium of claim 8 , wherein the error metric is defined as a squared error of the integerized filter coefficients of the filter candidate from the corresponding floating-point filter coefficients scaled by a particular value.

10 . The non-transitory computer-readable medium of claim 8 , wherein the error metric is calculated by approximating an integral of a spectral error between a first frequency response of the filter candidate and a second frequency response of an ideal interpolator, and
wherein the error metric is further defined by applying weights to the spectral error.

11 . The non-transitory computer-readable medium of claim 8 , wherein selecting the integerized interpolation filter from the set of filter candidates comprises:
selecting a reduced set of filter candidates from the set of filter candidates, each filter candidate in the reduced set of filter candidates having an error metric lower than remaining filter candidates in the set of filter candidates; and determine a rate-distortion result for each of the reduced set of filter candidates by applying respective filter candidates in a video coding system, wherein the integerized interpolation filter is selected from the reduced set of filter candidates based on the rate-distortion results.

12 . The non-transitory computer-readable medium of claim 8 , wherein two integerized filter coefficient values for each filter coefficient comprise a largest integer value that is smaller than the filter coefficient scaled by a particular value and a smallest integer value that is larger than the filter coefficient scaled by the particular value.

13 . The non-transitory computer-readable medium of claim 12 , wherein the set of filter candidates comprise filter candidates having filter coefficients selected from the two integerized filter coefficient values of respective filter coefficients.

14 . A method for decoding a video from a video bitstream, the method comprising:
decoding one or more frames of the video from the video bitstream; performing inter prediction based on the one or more frames using a set of integerized interpolation filters to decode another frame of the video, wherein the set of integerized interpolation filters are generated by: accessing a set of interpolation filters, each of the set of interpolation filters having floating-point filter coefficients; for each interpolation filter in the set of interpolation filters, generating two integerized filter coefficient values for each filter coefficient of the interpolation filter, generating a set of filter candidates based on the two integerized filter coefficient values for each filter coefficient, calculating an error metric for each filter candidate in the set of filter candidates, and selecting an integerized interpolation filter for the interpolation filter from the set of filter candidates, the selected integerized interpolation filter having a lowest error metric among the set of filter candidates; and causing the decoded one or more frame and the decoded another frame to be displayed.

15 . The method of claim 14 , wherein the error metric is defined as a squared error of the integerized filter coefficients of the filter candidate from the corresponding floating-point filte

Assignees

• Guangdong Oppo Mobile Telecommunications Corp Ltd

Inventors

• Gan Jonathan
• Yu Yue
• Yu Haoping

Classifications

• H04N19/52

  by predictive encoding · CPC title

• H04N19/184

  the unit being bits, e.g. of the compressed video stream · CPC title

• H04N19/172

  the region being a picture, frame or field · CPC title

• H04N19/136

  Incoming video signal characteristics or properties · CPC title

• H03H2017/022

  Rounding error · CPC title