Multi-Model Intra Prediction for Lower Chroma CCLM Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding methods face challenges in efficiently compressing digital video data due to high bandwidth demands and computational complexity, particularly in cross-component prediction techniques like the cross-component linear model (CCLM), which introduce issues such as increased memory requirements and complexity in hardware designs.
Innovation Solution
Simplified linear model derivations for cross-component prediction are implemented, reducing computational complexity and enhancing coding efficiency by optimizing parameter derivation methods for CCLM, including adaptive selection of chroma samples and luma samples based on position rules, and using simplified equations to derive α and β values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If cross-component linear model (CCLM) is used for chroma prediction, then prediction accuracy is improved, but computational complexity and memory requirements increase
Solution Approach 1:
The patent divides the chroma block into multiple regions and applies different prediction methods to each region. Specifically, it segments the chroma prediction process into: (1) regions where CCLM is applied with full complexity, (2) regions where simplified linear model is used, and (3) regions where DC prediction is sufficient. This segmentation allows the system to achieve high prediction accuracy in complex regions while reducing computational burden in simpler regions.
Solution Approach 2:
The patent introduces a parameter (e.g., chroma intra prediction mode flag, region classification parameters) that controls the complexity level of the prediction method applied. By changing these parameters based on block characteristics, the system can dynamically adjust between high-accuracy CCLM and low-complexity methods, resolving the contradiction between prediction accuracy and computational complexity.
2Measurement precision
If cross-component linear model (CCLM) is used for chroma prediction, then prediction accuracy is improved, but memory requirements increase
Solution Approach 1:
The patent extracts and removes unnecessary memory operations from the CCLM implementation. It does this by: (1) Reusing already-available luma and chroma samples from neighboring blocks without additional memory reads, (2) Eliminating redundant temporary variable storage, and (3) Computing prediction values using only the essential samples needed for the linear model, thereby reducing memory bandwidth and storage requirements while preserving prediction accuracy.
3Device complexity
If simplified linear model derivations are used, then computational complexity is reduced, but coding efficiency may be compromised
Solution Approach 1:
The patent implements a dynamic prediction strategy where the system adaptively selects between simplified linear model and full CCLM based on block characteristics. The decision is made dynamically using criteria such as block size, texture complexity, and gradient analysis. This dynamic approach ensures that simplified methods are used only when appropriate, maintaining coding efficiency while reducing overall computational complexity across the video stream.
Data Source
AI summary
A method for video processing is provided. The method includes determining, for a conversion between a current video block of a video that is a chroma block and a coded representation of the video, a final prediction P(x, y) of a chroma sample at a position (x, y) in the current video block as a combination of prediction results of multiple cross-component linear models (MCCLMs), wherein the MCCLMs are selected based on the position (x, y) of the chroma sample; and performing the conversion based on the final prediction.