Merge With MVD Mode Candidate Reordering for Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems face inefficiencies in processing camera-captured videos due to limited spatial correlation within the same picture, particularly in screen content coding, and the Merge with MVD Mode (MMVD) technique requires improvements in processing complexity and redundancy reduction.
Innovation Solution
A method and apparatus for reordering Motion Vector Prediction (MVP) candidates based on template matching costs and diversity conditions, moving similar candidates to later positions to reduce redundancy and improve coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If Merge with MVD Mode (MMVD) technique is used to improve coding efficiency, then compression performance is improved, but processing complexity increases
Solution Approach 1:
The patent reorders MVP candidates before the MMVD processing stage based on template matching costs and diversity conditions. By pre-arranging candidates in an optimal sequence that prioritizes diverse and promising options, the subsequent MMVD processing requires fewer iterations and operations to achieve the same coding efficiency, thus reducing processing complexity while maintaining compression performance
Solution Approach 2:
The patent changes the ordering parameter of MVP candidates from conventional sequences to sequences based on template matching costs and diversity metrics. This parameter transformation allows the system to identify and prioritize candidates that are more likely to yield better compression results with less processing, effectively reducing the complexity burden of MMVD while preserving coding efficiency
2Productivity
If traditional MVP candidate ordering is used, then processing is simpler, but coding efficiency is reduced
Solution Approach 1:
The patent transforms the candidate selection parameter from simple positional ordering to a composite metric based on template matching costs and diversity conditions. This parameter change enables more intelligent candidate prioritization that improves coding efficiency by presenting better candidates earlier in the processing sequence
Solution Approach 2:
The patent performs preliminary analysis of MVP candidates using template matching and diversity assessment before the main coding process. This preliminary action pre-sorts candidates in an optimal sequence, reducing the need for complex real-time evaluations during coding and achieving better efficiency without proportionally increasing overall complexity
3Measurement precision
If all MVP candidates are processed in MMVD, then coding accuracy is improved, but redundancy increases
Solution Approach 1:
The patent extracts and removes redundant MVP candidates from the processing sequence by reordering them to later positions based on diversity conditions. Candidates that are too similar to previously processed candidates are moved to lower priority positions, effectively filtering out redundancy while preserving the diverse, high-value candidates needed for accurate coding
Solution Approach 2:
The patent discards redundant candidates early in the reordered sequence by moving them to later positions where they are less likely to be selected. This selective discarding of redundant information maintains coding accuracy by focusing processing on diverse, informative candidates while reducing overall redundancy in the MMVD process
Data Source
AI summary
A method and apparatus for video coding are disclosed. A set of MVP candidates is determined for the current block. Member candidates in the set of MVP candidates are reordered according to template matching costs associated with the member candidates. A diversity condition is checked for a target MVP candidate pair in the reordered set, where the diversity condition comprises a target matching cost difference for the target MVP candidate pair being smaller than a first threshold from a first set of at least one pre-defined value, or a target MV difference for the target MVP candidate being smaller than a second threshold. If the diversity condition is satisfied, one of the target MVP candidate pair is moved to a later position in the reordered set of MVP candidates to form a modified-reordered set of MVP candidates. The current block is encoded or decoded by using motion information.