Granted Patent B2

US 12,388,985 · App. 18/733,595 · Granted Aug 12, 2025

Inventors: Xianglin Wang (San Diego, CA); Yi-Wen Chen (San Diego, CA); Xiaoyu Xiu (San Diego, CA); Tsung-Chuan Ma (San Diego, CA); Hong-Jheng Jhu (San Diego, CA); Shuiming Ye (San Diego, CA)

Assignee: BEIJING DAJIA INTERNET INFORMATION TECHNOLOGY CO., LTD.

H04N19/105H04N19/119H04N19/13H04N19/139H04N19/46H04N19/52H04N19/573

Abstract

Claims (40)

1. A method for video decoding with geometric prediction, comprising:

constructing, by a decoder, a first merge list comprising a plurality of candidates, based on a merge list construction process for regular merge prediction, wherein each one of the plurality of candidates is a motion vector comprising a List 0 motion vector, or a List 1 motion vector, or both;

receiving, by the decoder, a first index value to indicate a first candidate that is chosen from the first merge list;

receiving, by the decoder, a second index value to indicate a second candidate that is chosen from the first merge list;

receiving, by the decoder, a first binary flag that is coded using a first context-adaptive binary arithmetic coding (CABAC) context modeling method to indicate whether a List 0 motion vector of the first candidate or a List 1 motion vector of the first candidate is selected for a first PU of the geometric prediction; and

receiving, by the decoder, a second binary flag that is coded using a second CABAC context modeling method to indicate whether a List 0 motion vector of the second candidate or a List 1 motion vector of the second candidate is selected for a second PU of the geometric prediction,

wherein the first CABAC context modeling method or the second CABAC context modeling method comprises:

receiving each binary flag as a CABAC context bin, wherein a CABAC probability under each context model is initialized differently in response to whether a current picture uses a backward prediction.

2. The method for video decoding with geometric prediction of claim 1 , wherein at least two context models are available to decode the second binary flag, with a context model selected based on a value of the first binary flag.

3. The method for video decoding with geometric prediction of claim 1 , wherein the CABAC probability under each context model is initialized differently further comprises:

a higher CABAC initial probability is used for each binary flag to indicate selecting a List 0 motion vector in case the current picture does not use the backward prediction than in case the current picture uses the backward prediction.

4. The method for video decoding with geometric prediction of claim 1 , further comprising:

decoding, by the decoder, the first binary flag as a CABAC bypass bin, and decoding the second binary flag as a CABAC context coded bin.

5. An apparatus for video decoding with geometric prediction, comprising:

one or more processors; and

a memory configured to store instructions executable by the one or more processors;

wherein the one or more processors, upon execution of the instructions, are configured to:

construct a first merge list comprising a plurality of candidates, based on a merge list construction process for regular merge prediction, wherein each one of the plurality of candidates is a motion vector comprising a List 0 motion vector, or a List 1 motion vector, or both;

receive a first index value to indicate a first candidate that is chosen from the first merge list;

receive a second index value to indicate a second candidate that is chosen from the first merge list;

receive a first binary flag that is coded using a first context-adaptive binary arithmetic coding (CABAC) context modeling method to indicate whether a List 0 motion vector of the first candidate or a List 1 motion vector of the first candidate is selected for a first PU of the geometric prediction; and

receive a second binary flag that is coded using a second CABAC context modeling method to indicate whether a List 0 motion vector of the second candidate or a List 1 motion vector of the second candidate is selected for a second PU of the geometric prediction,

wherein the first CABAC context modeling method or the second CABAC context modeling method comprises:

receiving each binary flag as a CABAC context bin, wherein a CABAC probability under each context model is initialized differently in response to whether a current picture uses a backward prediction.

6. The apparatus for video decoding with geometric prediction of claim 5 , wherein at least two context models are available to decode the second binary flag, with a context model selected based on a value of the first binary flag.

7. The apparatus for video decoding with geometric prediction of claim 5 , wherein a higher CABAC initial probability is used for each binary flag to indicate selecting a List 0 motion vector in case the current picture does not use the backward prediction than in case the current picture uses the backward prediction.

8. The apparatus for video decoding with geometric prediction of claim 5 , wherein the one or more processors are further configured to:

decode the first binary flag as a CABAC bypass bin, and decode the second binary flag as a CABAC context coded bin.

9. A non-transitory computer-readable storage medium for video decoding stored computer-executable instructions and a bitstream that, when executed by one or more processors, cause the one or more processors to perform the following method with the bitstream:

constructing a first merge list comprising a plurality of candidates, based on a merge list construction process for regular merge prediction, wherein each one of the plurality of candidates is a motion vector comprising a List 0 motion vector, or a List 1 motion vector, or both;

receiving a first index value to indicate a first candidate that is chosen from the first merge list;

receiving a second index value to indicate a second candidate that is chosen from the first merge list;

receiving a first binary flag that is coded using a first context-adaptive binary arithmetic coding (CABAC) context modeling method to indicate whether a List 0 motion vector of the first candidate or a List 1 motion vector of the first candidate is selected for a first PU of the geometric prediction; and

receiving a second binary flag that is coded using a second CABAC context modeling method to indicate whether a List 0 motion vector of the second candidate or a List 1 motion vector of the second candidate is selected for a second PU of the geometric prediction,

wherein the first CABAC context modeling method or the second CABAC context modeling method comprises:

receiving each binary flag as a CABAC context bin, wherein a CABAC probability under each context model is initialized differently in response to whether a current picture uses a backward prediction.

10. The non-transitory computer-readable storage medium of claim 9 , wherein at least two context models are available to decode the second binary flag, with a context model selected based on a value of the first binary flag.

11. The non-transitory computer-readable storage medium of claim 9 , wherein a higher CABAC initial probability is used for each binary flag to indicate selecting a List 0 motion vector in case the current picture does not use the backward prediction than in case the current picture uses the backward prediction.

12. The non-transitory computer-readable storage medium of claim 9 , wherein the computer-executable instructions cause the one or more processors to further perform:

decoding the first binary flag as a CABAC bypass bin, and decoding the second binary flag as a CABAC context coded bin.

Assignments (1)

Continuity (4)

References Cited (14)