Chroma Residual Coding Using Cross-Component Modification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques fail to effectively reduce cross-component correlation between chroma components (Cb and Cr) in the YCbCr color space, leading to inefficiencies in bit usage and coding performance.
Innovation Solution
Applying residual modification functions (RMFs) to both Cb and Cr residual data, where each function modifies the respective chroma component based on the other, followed by forward and inverse transforms to encode and decode the modified data, thereby reducing the values and bits required for encoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional video coding techniques are used to encode chroma residual data, then the encoding process is simple, but the coding efficiency is suboptimal due to cross-component correlation between Cb and Cr components
Solution Approach 1:
The patent combines the encoding of Cb and Cr chroma residual data by applying residual modification functions that utilize cross-component correlation. Specifically, the Cb residual data is modified using Cr residual data and vice versa, merging the processing of both components to improve coding efficiency while managing complexity through structured function application.
Solution Approach 2:
The patent applies residual modification functions that transform the chroma residual data parameters. The first residual modification function modifies Cb residual data based on Cr residual data, and the second function modifies Cr residual data based on Cb residual data, changing the parameter representation to reduce bit usage and improve compression performance.
2Loss of information
If residual modification functions are applied to reduce cross-component correlation, then coding efficiency improves, but the computational complexity increases
Solution Approach 1:
The patent applies residual modification functions before the final encoding step. By pre-modifying the Cb and Cr residual data using cross-component information, the patent reduces the bit usage required for subsequent encoding, achieving better compression performance while managing computational complexity through advance processing.
Solution Approach 2:
The residual modification functions act as intermediaries between the original chroma residual data and the final encoded output. These functions use cross-component correlation as a mediator to transform the residual data, reducing bit usage efficiency loss while distributing computational complexity across structured transformation steps.
3Loss of energy
If separate encoding of Cb and Cr residual data is performed, then the encoding process is straightforward, but compression performance is suboptimal due to unexploited cross-component correlation
Solution Approach 1:
The patent merges the encoding processes for Cb and Cr residual data by applying modification functions that exploit cross-component correlation. The first function modifies Cb data using Cr data, and the second function modifies Cr data using Cb data, combining the processing to improve compression performance while maintaining manageable encoding complexity.
Solution Approach 2:
The patent transforms the chroma residual data parameters through residual modification functions that leverage cross-component relationships. By changing how Cb and Cr residual data are represented and processed, the patent achieves better compression performance while organizing the complexity into structured transformation operations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A device for coding video data includes a processor configured to generate an inter-prediction block and an intra-prediction block for a current block of video data; for each sample of a prediction block to be generated: determine a first weight for the sample according to a position of the sample in the prediction block; determine a second weight for the sample according to the position of the sample in the prediction block; apply the first weight to a sample at the position in the inter-prediction block to generate a weighted inter-prediction sample; apply the second weight to a sample at the position in the intra-prediction block to generate a weighted intra-prediction sample; and calculate a value for the sample at the position in the prediction block using the weighted inter-prediction sample and the weighted intra-prediction sample; and code the current block using the prediction block.