Motion Vector Candidate List Filtering for SbTMVP-MMVD Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies due to the lack of a redundant motion vector (MV) checking mechanism in subblock-based temporal motion vector prediction (SbTMVP), leading to duplicated MV fields and lack of MV diversity in candidate lists, which affects coding efficiency and hardware requirements.
Innovation Solution
Implement a checking mechanism to detect and remove redundant MVs, and adjust displacement vectors (DVs) to ensure they adhere to boundary conditions, enhancing MV diversity and reducing hardware buffer size needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If no redundant MV checking mechanism is implemented in SbTMVP, then candidate list construction is simpler and faster, but duplicated MV fields and lack of MV diversity occur, reducing coding efficiency
Solution Approach 1:
The patent applies preliminary action by implementing a redundant MV checking mechanism during candidate list construction that detects and removes duplicated MV fields before they affect coding efficiency. The checking mechanism proactively identifies duplicates using boundary conditions and MV field comparisons, preventing redundancy issues before they impact the coding process.
2Quantity of substance
If MMVD candidate position is not adjusted for boundary conditions, then candidate generation is simpler, but hardware buffer size must be larger to accommodate out-of-boundary values
Solution Approach 1:
The patent applies parameter changes by adjusting the MMVD candidate position parameters to satisfy boundary conditions. The displacement vector (DV) derivation is modified with boundary checks that constrain the candidate position within valid buffer boundaries, allowing smaller hardware buffers while maintaining correct candidate generation.
3Productivity
If redundant MVs are not removed from candidate list, then more candidates are available for selection, but MV diversity is reduced and coding efficiency deteriorates
Solution Approach 1:
The patent applies the taking out principle by extracting and removing redundant MV fields from the candidate list during construction. The checking mechanism identifies duplicated MVs and excludes them from the final candidate list, ensuring that only diverse and useful MV candidates remain, thereby maintaining MV diversity and improving coding efficiency.
Data Source
AI summary
An example method of video decoding includes receiving video data that includes a current block from a video bitstream. The method also includes generating a subblock-based motion vector prediction (SbTMVP) for a subblock of the current block and generating a SbTMVP-MMVD candidate by applying a merge motion vector difference (MMVD) to the SbTMVP. In accordance with a determination that a motion vector (MV) of the SbTMVP-MMVD candidate meets one or more criteria, the SbTMVP-MMVD candidate is inserted into a candidate list for the current block. In accordance with a determination that the MV of the SbTMVP-MMVD candidate does not meet the one or more criteria, the SbTMVP-MMVD candidate is not inserted into the candidate list for the current block. The method further includes reconstructing the current block using the candidate list.