Template-Matched Merge Candidate Reordering for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in predicting intra and inter block patterns, leading to suboptimal compression ratios and increased data requirements due to the use of less likely prediction directions and redundant motion vector coding.
Innovation Solution
Implement a method for video decoding that involves generating a list of merge candidates for a current block, dividing them into subgroups based on template matching costs, and reordering these subgroups to prioritize the most efficient merge candidates for reconstruction, utilizing spatial and temporal motion vector predictors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional merge candidate lists are used without reordering, then the decoding process is simpler, but the compression ratio is suboptimal due to less efficient prediction directions
Solution Approach 1:
The patent implements dynamic reordering of merge candidate subgroups based on template matching costs. The list order is not fixed but adapts according to the calculated costs for each candidate, allowing the system to optimize compression ratio by prioritizing better prediction directions while maintaining manageable complexity through structured grouping and cost-based sorting
Solution Approach 2:
The patent changes the ordering parameter of merge candidates from a fixed sequential order to a cost-based order determined by template matching. By calculating and comparing template matching costs for each candidate subgroup, the system transforms the static candidate list into a dynamically optimized sequence that improves compression efficiency
2Measurement precision
If more merge candidates are included in the list, then the prediction accuracy improves, but the data redundancy increases
Solution Approach 1:
The patent extracts and separates merge candidates into distinct subgroups based on their source (spatial neighbors, temporal references, non-adjacent positions). This extraction allows the system to evaluate and prioritize candidates from different sources independently, selecting only the most relevant candidates for the final ordered list, thereby improving prediction accuracy while reducing redundancy
Solution Approach 2:
The patent implicitly discards less efficient merge candidates by placing them lower in the reordered list or excluding them from consideration. By evaluating template matching costs and ordering candidates accordingly, the system recovers only the most valuable prediction information, eliminating redundant data while maintaining prediction accuracy
3Manufacturing precision
If template matching is performed for all merge candidates, then the prediction quality improves, but the computational complexity increases
Solution Approach 1:
The patent segments the merge candidate evaluation process by dividing candidates into subgroups and performing template matching within each subgroup separately. This segmentation allows the system to manage computational complexity by processing candidates in manageable batches while still achieving high prediction quality through comprehensive evaluation of all subgroups
Solution Approach 2:
The patent applies template matching selectively to merge candidate subgroups rather than uniformly to all candidates. By performing partial template matching on prioritized subgroups first and using the results to guide further processing, the system achieves high prediction quality while controlling computational complexity through staged evaluation
Data Source
AI summary
In a method, coded information of a current block and neighboring blocks of the current block in a current picture is received from a coded video bitstream. A list of merge candidates of the current block is generated based on the neighboring blocks of the current block. The list of merge candidates of the current block is divided into a plurality of subgroups. Each of the plurality of subgroups includes one or more merge candidates. The one or more merge candidates are ordered within each subgroup by a respective template matching (TM) cost associated with each of the one or more merge candidates. The current block is reconstructed based on a merge candidate selected from the list of merge candidates of the current block.