Template-Based Video Prediction Using Adaptive Template Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques struggle to efficiently utilize neighboring samples for template matching-based prediction, leading to suboptimal video coding efficiency and quality, especially with increasing image sizes and resolutions.
Innovation Solution
A method and apparatus that set an adjacent or non-adjacent region as a template based on neighboring reconstructed regions, using template matching to enhance video coding efficiency and quality by determining a distance index and template region shape for optimal template search within the reconstructed region.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If template matching-based prediction is performed using adjacent reconstructed samples, then video coding efficiency and quality are improved, but the complexity of determining the optimal template region increases
Solution Approach 1:
The patent applies parameter changes by introducing a distance index (adjacent_idx) to control the distance between the current block and template region, and by adjusting the shape parameter (tmShape) of the template region. These parameters allow flexible adaptation of the template search strategy to different coding scenarios, resolving the contradiction between improving coding efficiency and managing determination complexity.
Solution Approach 2:
The patent implements dynamics by making the template region configuration adaptive rather than fixed. The template region's distance and shape are dynamically determined based on the distance index and shape index from the bitstream, allowing the system to optimize template matching performance for different blocks and coding conditions while maintaining manageable complexity through standardized adaptation rules.
2Measurement precision
If the template region is set based on distance index and shape index, then the accuracy of template matching is improved, but the amount of information to be decoded increases
Solution Approach 1:
The patent applies partial action by selectively transmitting distance index and shape index information only when template matching-based prediction is enabled and when such information is beneficial for the current block. This partial information transmission approach improves template matching accuracy for blocks that benefit from it while minimizing the overall bitstream overhead by avoiding unnecessary information transmission for other blocks.
Data Source
AI summary
A method and an apparatus are disclosed for video coding using template-based prediction. In the disclosed embodiments, a video decoding device decodes a distance index (adjacent_idx) that indicates a distance between the current block and a template region. The video decoding device determines the template region based on the distance between the current block and the template region, and a shape of the template region. The video decoding searches for an optimal template in a search region by using the template region and based on template matching and generates a prediction block of the current block from a reference block corresponding to the optimal template.