Mode-Dependent JCST to Reduce Chroma Coding Redundancy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding formats like AV1 face inefficiencies in residual coding due to high correlation between chroma channels (Cb and Cr), leading to sub-optimal joint chroma component transforms that do not effectively handle all prediction modes, resulting in redundancy and reduced coding efficiency.
Innovation Solution
Implementing a mode-dependent joint component secondary transformation (JCST) that adapts the transform kernel based on prediction modes, intra/inter prediction types, and primary/secondary transform types to optimize coding efficiency by reducing statistical redundancy between chroma components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a fixed joint chroma component transform is used, then the transform is simple to implement, but it does not effectively handle all prediction modes resulting in redundancy and reduced coding efficiency
Solution Approach 1:
The patent applies dynamics by making the transform kernel adaptive rather than fixed. Different transform kernels are selected based on prediction mode (intra/inter), chroma type (Cb/Cr), and block size. This allows the system to optimize coding efficiency for each specific condition while maintaining manageable complexity through structured selection criteria.
Solution Approach 2:
The patent changes the parameter of transform kernel selection based on multiple conditions including prediction mode, chroma component type, and block size. By varying these parameters dynamically, the system achieves optimal compression performance for different video characteristics without requiring a completely complex redesign of the transformation architecture.
2Productivity
If chroma channels are processed independently, then the processing is simple, but the high correlation between Cb and Cr channels is not exploited resulting in redundant information
Solution Approach 1:
The patent merges the processing of Cb and Cr chroma channels by applying joint transformation. Instead of processing channels independently, the system transforms them together, exploiting the high correlation between channels to reduce redundancy. This is achieved through joint chroma component transformation that operates on both channels simultaneously.
Solution Approach 2:
The patent creates a universal joint transform structure that handles different chroma channel combinations (Cb-Cr, Cr-Cb) and different prediction modes through a single flexible framework. The transform kernel selection mechanism provides multi-functionality across various coding conditions while maintaining a consistent processing approach.
3Productivity
If a single transform kernel is used for all prediction modes, then the implementation is simple, but coding efficiency is reduced due to sub-optimal transforms for specific modes
Solution Approach 1:
The patent applies local quality by selecting different transform kernels tailored to specific local conditions (prediction modes, chroma types, block sizes). Instead of using a single universal kernel, the system adapts the transform characteristics to match the specific coding conditions, optimizing compression performance for each local scenario.
Solution Approach 2:
The patent implements dynamics through conditional transform kernel selection. The transform choice is not static but dynamically adjusts based on prediction mode (intra/inter), chroma component (Cb/Cr), and block size. This dynamic adaptation enables optimal coding efficiency across diverse video content while maintaining systematic complexity management.
Data Source
AI summary
A method for coding video data, executable by a processor, includes receiving video data; entropy-parsing the received video data into one or more components; de-quantizing the one or more entropy-parsed components; performing a joint component secondary transformation (JCST) on the one or more components in accordance with a JCST kernel selected based on one of (i) a prediction mode corresponding to the video data, (ii) a primary transform type of a current block, and (iii) a secondary transform kernel selected for the current block; and decoding the video data based on one or more residual components corresponding to the joint component secondary transformed components.