Intra Template Matching Search Region Selection for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding methods using Intra Template Matching Prediction (Intra TMP) require more effective reference information to achieve higher prediction accuracy, leading to increased encoding complexity and reduced efficiency.
Innovation Solution
The method involves determining a first template for a current block and selecting a fully reconstructed and/or undetermined reconstructed search region, limiting the search to a smaller selected region to determine Block Vectors (BVs), thereby reducing encoding and decoding complexity while utilizing available samples effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If more effective reference information is acquired to achieve higher prediction accuracy, then prediction accuracy is improved, but encoding complexity increases
Solution Approach 1:
The search region is divided into multiple sub-regions, and template matching is performed separately in each sub-region. This segmentation allows the algorithm to explore more reference information systematically while managing computational complexity through divided processing steps.
Solution Approach 2:
The patent implements dynamic adjustment of search region size and configuration based on template characteristics and prediction needs. The search region is adaptively expanded or contracted, and sub-region boundaries are dynamically determined, allowing the system to acquire more reference information when needed while reducing complexity when sufficient accuracy is already achieved.
2Measurement precision
If more effective reference information is acquired to achieve higher prediction accuracy, then prediction accuracy is improved, but encoding and decoding efficiency decreases
Solution Approach 1:
By dividing the search into multiple sub-regions with different processing strategies, the patent enables parallel processing and optimizes computational resource allocation. This segmentation improves efficiency by avoiding exhaustive search across the entire region while still accessing sufficient reference information for accurate prediction.
Solution Approach 2:
The patent performs template matching in selected sub-regions rather than the complete search region, applying partial action principle. This approach processes only the necessary portions of the reference information required to achieve adequate prediction accuracy, thereby maintaining efficiency while still acquiring sufficient effective reference information.
Data Source
AI summary
The embodiments of the present application disclose an encoding method, a decoding method, and a storage medium. The encoding/decoding method includes: an encoder/decoder determining a first template corresponding to the current block; according to the first template, determining a fully reconstructed search area and/or an undetermined reconstructed search area, wherein the fully reconstructed search area comprises reconstructed sampling points, and the undetermined reconstructed search area comprises reconstructed sampling points and/or unreconstructed sampling points; performing area selection on the basis of the fully reconstructed search area and/or the undetermined reconstructed search area, and determining a selected fully reconstructed search area and/or a selected undetermined reconstructed search area; and respectively searching in the selected fully reconstructed search area and/or the selected undetermined reconstructed search area to determine one or more block vectors of the current block.