Three-Mode Template Intra Coding for Smooth Video Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies struggle to achieve high compression efficiency while maintaining video quality, particularly in blocks with smooth texture/content, due to limitations in intra prediction mode derivation.
Innovation Solution
Populating a candidate list for a current block with three-mode entries, including non-angular modes like DC mode, planar mode, or intra block copy mode, to enhance intra prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional intra prediction mode derivation is used, then device complexity is reduced, but coding accuracy deteriorates particularly for blocks with smooth texture/content
Solution Approach 1:
The candidate list is segmented into multiple types of candidates (angular modes, non-angular modes, and IBC modes) rather than using a single uniform approach. This segmentation allows the system to select appropriate prediction modes based on block characteristics, improving coding accuracy for different texture types while managing complexity through structured organization.
Solution Approach 2:
The candidate list construction is made dynamic by adaptively selecting and populating candidates based on neighboring block configurations and block characteristics. The system dynamically adjusts the composition of the candidate list rather than using a fixed structure, enabling improved accuracy for smooth texture blocks while maintaining efficiency through conditional logic.
2Measurement precision
If more prediction modes are considered, then coding accuracy improves, but processing time increases
Solution Approach 1:
The candidate list is preliminarily constructed and populated with potential prediction modes before the actual prediction process. By pre-organizing angular, non-angular, and IBC candidates based on neighboring block information, the system reduces the computational burden during decoding/encoding while maintaining access to multiple prediction options for improved accuracy.
Solution Approach 2:
The system constructs a candidate list with a limited number of carefully selected candidates rather than evaluating all possible prediction modes. This partial action approach includes only the most relevant angular, non-angular, and IBC candidates, providing sufficient prediction accuracy for smooth texture blocks without the computational cost of exhaustive mode evaluation.
3Measurement precision
If three-mode entries including non-angular modes are used, then coding accuracy for smooth texture blocks improves, but bitstream complexity increases
Solution Approach 1:
The candidate list construction applies local quality by including non-angular modes (DC mode, planar mode) and IBC modes specifically for blocks where these modes are beneficial, such as smooth texture blocks. Rather than uniformly applying all modes to all blocks, the system locally adapts the candidate composition based on block characteristics, improving accuracy where needed while avoiding unnecessary complexity elsewhere.
Solution Approach 2:
The candidate list structure is designed to be universal by incorporating multiple types of prediction modes (angular, non-angular, and IBC) that can handle different block characteristics. This multi-functional candidate list can serve various block types within a unified framework, improving coding accuracy across different scenarios without requiring separate processing paths that would increase bitstream complexity.
Data Source
AI summary
The various implementations described herein include methods and systems for coding video. An example method of video decoding includes receiving a video bitstream comprising a plurality of blocks that includes a current block. The method also includes, for each neighboring block in a set of neighboring blocks of the current block, when the neighboring block is coded using a template-based intra prediction mode derivation (TIMD), populating a candidate list for the current block with an entry for the neighboring block, where at least one entry in the candidate list comprises a three-mode entry. The method further includes reconstructing the current block using an entry from the candidate list.