Template Matching Refinement for CIIP and GPM Video Modes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies do not enable template matching refinement for certain inter-prediction modes like combined inter/intra prediction (CIIP), geometric partitioning mode (GPM), and multi-hypothesis prediction (MHP), limiting video coding gains in these modes.
Innovation Solution
The implementation of template matching refinement techniques to generate motion vectors and prediction blocks for blocks inter-predicted in CIIP, GPM, and MHP modes, allowing for improved prediction accuracy by comparing reference templates within a search area to find the closest match.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If template matching refinement is not applied to CIIP, GPM, and MHP modes, then device complexity remains low, but video coding efficiency and prediction accuracy are limited
Solution Approach 1:
The patent applies template matching refinement separately to different inter-prediction modes (CIIP, GPM, MHP) by enabling it selectively based on mode type. This segmentation allows the complex template matching process to be applied only where beneficial rather than universally, improving video coding efficiency for specific modes while avoiding unnecessary complexity in others.
2Measurement precision
If template matching is applied to generate motion vectors for all inter-prediction modes, then prediction accuracy improves, but processing time and computational load increase
Solution Approach 1:
The patent dynamically adjusts the application of template matching refinement based on the inter-prediction mode being used. The system enables template matching for modes that benefit from it (CIIP, GPM, MHP) while allowing other modes to use standard motion vector generation, creating a dynamic processing approach that balances accuracy with processing time requirements.
3Productivity
If template matching refinement is enabled for CIIP, GPM, and MHP modes, then video coding gains are extended to these modes, but computational complexity increases
Solution Approach 1:
The patent applies template matching refinement locally to specific inter-prediction modes (CIIP, GPM, MHP) rather than uniformly across all modes. This local quality approach enables video coding gains to be extended to modes that can utilize them effectively, while avoiding the computational complexity overhead in modes where template matching provides less benefit.
Data Source
AI summary
A device for decoding video data includes memory configured to store the video data and processing circuitry. The processing circuitry is configured to determine that a current block of the video data is inter-predicted in a combined inter-intra prediction (CIIP) mode or a geometric partitioning mode (GPM), determine that template matching is enabled for the current block, generate a motion vector for the current block based on template matching; determine a prediction block for the current block based on the motion vector in accordance with the CHIP mode or the GPM, and reconstruct the current block based on the prediction block.