Adaptive Colour Transform in Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding techniques, such as HEVC, face inefficiencies in encoding camera-captured video content due to difficulties in finding exact matches for Current Picture Referencing (CPR) within the same picture, especially with gradual spatial changes, and the Adaptive Colour Transform (ACT) is not effectively applied when luma and chroma components are coded using different splitting trees.
Innovation Solution
The implementation of Adaptive Colour Transform (ACT) is restricted to blocks coded using single tree block partitioning, with the ACT flag being disabled for dual tree block partitioning, and the colour space conversion is performed only when the luma and chroma components are coded using the same splitting trees, ensuring effective colour transform operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If Adaptive Colour Transform is applied to blocks with dual tree block partitioning, then coding flexibility is improved, but colour transform accuracy deteriorates due to mismatched luma and chroma block sizes
Solution Approach 1:
The patent applies different processing rules to different block partitioning types. For single tree block partitioning, ACT is fully applied to both luma and chroma components. For dual tree block partitioning, ACT is disabled or selectively applied only to luma components. This local differentiation ensures transform accuracy is maintained where block sizes match, while still providing coding flexibility where partial application is beneficial.
2Adaptability or versatility
If ACT is applied to all block types, then coding versatility is improved, but prediction accuracy deteriorates due to colour space conversion errors in dual tree partitions
Solution Approach 1:
The patent implements local quality by adapting the ACT application strategy to the specific block partitioning type. In regions with single tree partitioning where colour space conversion is accurate, ACT is applied to both luma and chroma. In regions with dual tree partitioning where conversion errors occur, ACT is restricted to luma only or disabled entirely, thereby maintaining prediction accuracy while preserving coding versatility in suitable regions.
3Adaptability or versatility
If colour space conversion is performed for dual tree block partitioning, then colour coding flexibility is improved, but transform errors increase due to size mismatches between luma and chroma blocks
Solution Approach 1:
The patent applies local quality principles by performing colour space conversion only for blocks where luma and chroma components share the same partitioning tree. For single tree partitions, full YCbCr conversion is performed. For dual tree partitions, conversion is either skipped or limited to luma components only, avoiding the transform errors that would arise from size mismatches between luma and chroma blocks.
Data Source
AI summary
Video processing methods include receiving input data of a current block in a current picture, determining if a luma component and chroma components of the current block are coded using different splitting trees, disabling Adaptive Colour Transform (ACT) on the current block when the luma component and the chroma components of the current block are coded using different splitting trees, and encoding or decoding the current block. ACT is only allowed to be performed on blocks coded in single tree block partitioning to convert a RGB colour space into an YCoCg colour space. ACT is applied to residues according to some embodiments, or ACT is applied to input data and predictors according to some other embodiments.