Adaptive Merge Candidate Reordering with Zero Motion Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in handling merge candidates with zero motion vectors, leading to suboptimal video encoding and decoding performance.
Innovation Solution
The introduction of adaptive reordering of merge candidates (ARMC) with template matching (TM) to include and reorder merge candidates, including those with zero motion vectors, to enhance video encoding and decoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If merge candidates with zero motion vectors are excluded from the merge candidate list, then the encoding complexity is reduced, but the video encoding efficiency and compression quality deteriorate
Solution Approach 1:
The patent applies preliminary action by pre-calculating template matching costs for all merge candidates including those with zero motion vectors, and pre-ordering them in the merge candidate list before actual encoding. This allows the decoder to efficiently select the best candidate without complex runtime calculations, resolving the contradiction between including zero MV candidates (improves efficiency) and maintaining low complexity.
Solution Approach 2:
The patent changes the ordering parameter of merge candidates based on template matching costs rather than traditional spatial or temporal order. By reordering candidates with lower template matching costs to higher priority positions, the system improves encoding efficiency while maintaining manageable complexity through a systematic reparameterization of the candidate selection process.
2Manufacturing precision
If adaptive reordering of merge candidates with template matching is implemented, then the compression quality is improved, but the processing time and computational load increase
Solution Approach 1:
The patent performs template matching and cost calculation in advance during the merge candidate list construction phase, so that when actual encoding occurs, the reordering is already complete. This preliminary computation approach improves compression quality through accurate template matching while minimizing the time penalty during the critical encoding path.
Solution Approach 2:
The system uses the template matching mechanism to automatically evaluate and reorder merge candidates based on their actual matching performance with the current block. This self-service approach allows the data to determine its own optimal ordering based on content characteristics, improving compression quality without requiring manual intervention or complex external control.
3Measurement precision
If zero motion vector candidates are added to the merge candidate list, then the accuracy of motion prediction is improved, but the number of candidates to process increases
Solution Approach 1:
The patent applies local quality by adding zero motion vector candidates specifically at positions in the merge candidate list where they are most beneficial - after reordering based on template matching costs. Rather than uniformly processing all candidates, the system places zero MV candidates in optimal local positions within the list, improving prediction accuracy for stationary regions while keeping the total processing load manageable through selective placement.
Data Source
AI summary
Aspects of the disclosure provide a method and an apparatus including processing circuitry that adds merge candidates into a merge candidate list of a current block in a current picture. The merge candidates include available merge candidates including one or more of spatial motion vector predictors (MVP(s)), temporal MVP(s), non-adjacent MVP(s), history-based MVP(s), and pairwise average MVP(s). Whether the merge candidate list includes all available SMVP, TMVP, NA-MVP, HMVP, and PAMVP, and no merge candidate in the merge candidate list has a zero motion vector (zero MV) are determined. If the merge candidate list includes all available SMVP, TMVP, NA-MVP, HMVP, and PAMVP, and no merge candidate in the merge candidate list has the zero MV, the merge candidate having the zero MV is added to the merge candidate list. Merge candidates in the merge candidate list including the merge candidate having the zero MV are reordered using template matching.