CCLM Chroma Prediction Using Outlier-Robust Luma Pair Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The simplified cross-component linear model (CCLM) fitting algorithm in image and video coding is susceptible to statistical outliers, leading to inaccurate line-fittings and increased chromatic coding inefficiencies, particularly in scenarios with strong variances in extreme luma values.
Innovation Solution
A decoder and encoder are configured to determine two sets of luma samples in a neighborhood of a current block, fitting a straight-line through pairs of luma-chroma samples from these sets, and using this linear function to predict chroma samples, thereby reducing susceptibility to statistical outliers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simplified CCLM fitting algorithm is used (fitting line between minimum and maximum luma value points), then computational complexity is reduced and decoding runtime is decreased, but susceptibility to statistical outliers increases leading to inaccurate line-fittings
Solution Approach 1:
The patent applies partial action by using a subset of luma-chroma sample pairs (specifically, pairs formed from extreme luma values and their corresponding chroma values) rather than all available samples. This selective approach reduces computational complexity while maintaining improved robustness through the specific choice of extreme value pairs that are less susceptible to outliers.
2Device complexity
If simplified CCLM fitting algorithm is used, then number of operations is reduced, but chromatic coding efficiency deteriorates due to increased susceptibility to statistical outliers
Solution Approach 1:
The patent changes the parameters used in line-fitting from all luma-chroma sample pairs to specifically selected pairs formed by extreme luma values and their corresponding chroma values. This parameter change transforms the fitting process to be more robust against outliers while maintaining low computational complexity, thereby improving chromatic coding efficiency without significantly increasing device complexity.
3Measurement precision
If LMS-based CCLM is used, then line-fitting accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the set of all luma-chroma sample pairs into specific subsets based on extreme luma values. By dividing the data into meaningful segments (minimum luma pairs and maximum luma pairs) and using only these segments for fitting, the method achieves robustness comparable to or better than LMS while significantly reducing computational complexity by avoiding processing of all samples.
Data Source
AI summary
A decoder for block-based decoding of picture data of a picture is configured to decode a current block of the picture by determining first and second sets of luma in a neighborhood of the current block, the first set of luma samples having smaller luma values than the second set of luma samples, fitting a straight-line having a linear function through first and second pairs of luma-chroma samples, wherein the first and second pairs of luma-chroma samples are determined based on the first and second sets of luma samples, respectively, and on first and second corresponding chroma samples, respectively, in said neighborhood of the current block, and predicting chroma samples of the current block using luma samples of the current block and said linear function. The present invention further concerns a respective encoder and corresponding methods for encoding and decoding.