Video Coder MIP and CCLM Prediction Harmonization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding and decoding technologies face complexity and inefficiency due to the need for separate down-sampling techniques for matrix intra prediction (MIP) and cross-component linear model (CCLM) prediction, which increases computational burden and reduces accuracy.
Innovation Solution
Configuring video coders to use down-sampled luma neighboring samples generated during MIP for both MIP and CCLM prediction, thereby simplifying the process and reducing complexity by applying a single down-sampling function for both techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If separate down-sampling techniques are used for MIP and CCLM prediction, then each prediction method can be optimized independently, but the computational complexity and device complexity increase
Solution Approach 1:
The patent merges the down-sampling operations for MIP and CCLM prediction into a single shared operation. The video coder performs one down-sampling of luma neighboring samples that serves both the MIP prediction of luma samples and the CCLM prediction of chroma samples, eliminating redundant computations while maintaining prediction accuracy for both functions.
Solution Approach 2:
The down-sampled luma neighboring samples are designed to serve multiple functions simultaneously: they are used both for MIP prediction of luma samples and for CCLM prediction of chroma samples. This multi-functional use of the same processed data reduces the overall computational burden without sacrificing the precision of either prediction method.
2Measurement precision
If separate down-sampling techniques are performed for MIP and CCLM prediction, then each prediction can be optimized, but the processing time and productivity decrease
Solution Approach 1:
The patent combines the down-sampling processing into a single operation that produces results used by both MIP and CCLM prediction pipelines. This merging eliminates the need to perform down-sampling twice, directly reducing processing time while maintaining the accuracy benefits of dedicated down-sampling for each prediction method.
Solution Approach 2:
The down-sampling of luma neighboring samples is performed once in advance as a preliminary step, and the resulting down-sampled samples are then reused for both MIP and CCLM prediction. This preliminary action avoids redundant processing and accelerates the overall encoding/decoding process while preserving prediction quality.
3Device complexity
If down-sampled luma neighboring samples are shared between MIP and CCLM prediction, then computational complexity is reduced, but the adaptability for different prediction requirements may be limited
Solution Approach 1:
The patent designs the down-sampled luma neighboring samples to be universally applicable for both MIP and CCLM prediction. The samples are processed once and then adaptively used in both prediction methods, demonstrating that a single processing pipeline can serve multiple prediction requirements without sacrificing the specific needs of each method.
Data Source
AI summary
A method of decoding video data includes predicting luma samples for a block of the video data using matrix intra prediction (MIP), where using MIP comprises down-sampling a set of luma neighboring samples to generate down-sampled luma neighboring samples. The method further includes predicting chroma samples for the block using cross-component linear model (CCLM) prediction, where using CCLM prediction comprises predicting the chroma samples for the block based on the down-sampled luma neighboring samples generated from the MIP. The method further includes determining a prediction block for the block based on the luma samples and the chroma samples, decoding a residual block for the block, and combining the prediction block and the residual block to decode the block.