Granted Patent B2

US 12,368,880 · App. 18/156,666 · Granted Jul 22, 2025

Inventors: Li Zhang (San Diego, CA); Kai Zhang (San Diego, CA); Hongbin Liu (Beijing, CN); Yue Wang (Beijing, CN)

Assignees: BEIJING BYTEDANCE NETWORK TECHNOLOGY CO., LTD.; BYTEDANCE INC.

H04N19/52H04N19/176

Abstract

Claims (53)

1. A video processing method, comprising:

maintaining one or more motion candidate tables, wherein each motion candidate table comprises one or multiple motion candidates, at least one of which is derived from at least one previous video block, and arrangement of the motion candidates in the motion candidate table is based on a sequence of addition of the motion candidates into the motion candidate table;

performing a motion candidate list derivation process to derive a motion candidate list for a first video block;

deriving, based on the motion candidate list, motion information for the first video block;

refining the motion information using a merge with a motion vector differences tool;

performing a conversion between the first video block and a bitstream of a video comprising the first video block based on the refined motion information; and

updating a motion candidate table of the one or more motion candidate tables by using the refined motion information refined by the merge with the motion vector differences tool;

wherein the one or more motion candidates in the updated motion candidate table are checked in an order during constructing a motion candidate list for a subsequent block.

2. The method of claim 1 , wherein the first video block is coded with a merge mode.

3. The method of claim 1 , wherein the motion candidate table is determined to be updated based on a coding mode of a current block belonging to at least one specific mode.

4. The method of claim 3 , wherein the at least one specific mode comprises at least one of a motion vector prediction mode or a merge mode.

5. The method of claim 3 , wherein the at least one specific mode excludes at least one of a sub-block merge mode or a triangular merge mode, wherein the motion candidate table is not updated using motion information derived from blocks using the sub-block merge mode or the triangular merge mode.

6. The method of claim 1 , wherein the motion candidate table is a history motion vector prediction (HMVP) table.

7. The method of claim 1 , wherein the updating comprises:

updating the motion candidate table using a pruning operation, wherein the pruning operation comprises comparing a new motion candidate to be added in the motion candidate table with at least one motion candidate in the motion candidate table to determine whether to delete the at least one motion candidate from the motion candidate table.

8. The method of claim 1 , wherein when a new candidate associated with motion information derived from the motion candidate list is a motion candidate added into the motion candidate list from the motion candidate table, the motion candidate table is reordered without a pruning operation.

9. The method of claim 1 , wherein when a new candidate associated with the motion information derived from the motion candidate list is a motion candidate related to a spatial neighboring block in a current picture comprising the first video block or a temporal block in a picture different from the current picture, the motion candidate table is updated with a pruning operation.

10. The method of claim 1 , further comprising:

performing a motion candidate list derivation process to derive a motion candidate list for a second video block which is coded with a motion vector prediction mode;

deriving, based on the motion candidate list derived for the second video block, motion information for the second video block;

performing a conversion between the second video block and a bitstream of a video comprising the second video block based on the motion information derived for the second video block; and

updating the motion candidate table using the motion information derived for the second video block without refinement.

11. The method of claim 1 , wherein the performing of the conversion includes generating the bitstream based on the first video block.

12. The method of claim 1 , wherein the performing of the conversion includes generating the first video block from the bitstream.

13. An apparatus for processing video data comprising a processor and a non-transitory memory with instructions thereon, wherein the instructions upon execution by the processor, cause the processor to:

maintain one or more motion candidate tables, wherein each motion candidate table comprises one or multiple motion candidates, at least one of which is derived from at least one previous video block, and arrangement of the motion candidates in the motion candidate table is based on a sequence of addition of the motion candidates into the motion candidate table;

perform a motion candidate list derivation process to derive a motion candidate list for a first video block;

derive, based on the motion candidate list, motion information for the first video block;

refine the motion information using a merge with a motion vector differences tool;

perform a conversion between the first video block and a bitstream of a video comprising the first video block based on the refined motion information; and

update a motion candidate table of the one or more motion candidate tables by using the refined motion information refined by the merge with the motion vector differences tool;

wherein the one or more motion candidates in the updated motion candidate table are checked in an order during constructing a motion candidate list for a subsequent block.

14. The apparatus of claim 13 , wherein the first video block is coded with a merge mode.

15. The apparatus of claim 13 , wherein the motion candidate table is determined to be updated based on a coding mode of a current block belonging to at least one specific mode.

16. The apparatus of claim 15 , wherein the at least one specific mode comprises at least one of a motion vector prediction mode or a merge mode.

17. The apparatus of claim 15 , wherein the at least one specific mode excludes at least one of a sub-block merge mode or a triangular merge mode, wherein the motion candidate table is not updated using motion information derived from blocks using the sub-block merge mode or the triangular merge mode.

18. The apparatus of claim 13 , wherein the motion candidate table is a history motion vector prediction (HMVP) table.

19. A non-transitory computer-readable storage medium storing instructions that cause a processor to:

maintain one or more motion candidate tables, wherein each motion candidate table comprises one or multiple motion candidates, at least one of which is derived from at least one previous video block, and arrangement of the motion candidates in the motion candidate table is based on a sequence of addition of the motion candidates into the motion candidate table;

perform a motion candidate list derivation process to derive a motion candidate list for a first video block;

derive, based on the motion candidate list, motion information for the first video block;

refine the motion information using a merge with a motion vector differences tool;

perform a conversion between the first video block and a bitstream of a video comprising the first video block based on the refined motion information; and

update a motion candidate table of the one or more motion candidate tables by using the refined motion information refined by the merge with the motion vector differences tool;

wherein the one or more motion candidates in the updated motion candidate table are checked in an order during constructing a motion candidate list for a subsequent block.

20. A non-transitory computer-readable recording medium storing a bitstream of a video which is generated by a method performed by a video processing apparatus, wherein the method comprises:

maintaining one or more motion candidate tables, wherein each motion candidate table comprises one or multiple motion candidates, at least one of which is derived from at least one previous video block, and arrangement of the motion candidates in the motion candidate table is based on a sequence of addition of the motion candidates into the motion candidate table;

performing a motion candidate list derivation process to derive a motion candidate list for a first video block;

deriving, based on the motion candidate list, motion information for the first video block;

refining the motion information using a merge with a motion vector differences tool;

generating the bitstream based on the refined motion information; and

updating a motion candidate table of the one or more motion candidate tables by using the refined motion information refined by the merge with the motion vector differences tool;

wherein the one or more motion candidates in the updated motion candidate table are checked in an order during constructing a motion candidate list for a subsequent block.

Assignments (2)

Priority Claims (1)

Continuity (3)

References Cited (400)

Cited By (3)

US 12,549,756 US 12,604,029 US 12,641,216