Luma-Guided Chroma Prediction for Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently compressing video data while maintaining quality due to limited bandwidth and memory resources, particularly in handling the correlation between luma and chroma components.
Innovation Solution
Implementing cross-component prediction methods such as Edge-classified Linear Model (ELM), Filter-based Linear Model (FLM), Gradient Linear Model (GLM), Convolutional Cross-Component Model (CCCM), and Multiple Linear Regression (MLR) models to predict chroma samples based on classified luma samples, utilizing various linear prediction techniques and pseudo-inverse matrix calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional video coding methods are used, then bandwidth and memory resources are limited, but video quality and compression efficiency deteriorate
Solution Approach 1:
The patent changes the parameters of prediction by using multiple linear regression models with different coefficients (α and β) to predict chroma samples from luma samples. This allows more flexible and accurate prediction compared to traditional fixed-parameter methods, improving compression efficiency while reducing the bandwidth needed to transmit video data.
Solution Approach 2:
The patent replaces traditional mechanical down-sampling processes with linear prediction models that compute chroma samples mathematically from luma samples. This substitution eliminates the need for physical down-sampling operations while maintaining or improving prediction accuracy, thereby reducing resource consumption.
2Loss of information
If down-sampling processes are applied to reduce data size, then bandwidth consumption decreases, but prediction accuracy and video quality deteriorate
Solution Approach 1:
The patent uses multiple linear regression models with optimizable parameters (α and β coefficients) to maintain high prediction accuracy without down-sampling. By adjusting these parameters based on the specific video content and block characteristics, the system achieves accurate chroma prediction while working with full-resolution luma samples, avoiding information loss.
Solution Approach 2:
The patent performs preliminary classification of luma blocks into different types (e.g., flat, gradient, edge) before applying appropriate linear prediction models. This preliminary action allows the system to select the most suitable prediction strategy for each block, improving overall prediction accuracy without needing to down-sample the data.
3Manufacturing precision
If complex prediction models are used to improve chroma sample prediction, then video quality improves, but computational complexity and processing time increase
Solution Approach 1:
The patent segments the video data into luma and chroma components and further divides luma blocks into different types based on their characteristics. By segmenting the processing task, the system can apply simple linear regression models to each segment rather than using a single complex model for all data, reducing overall computational complexity while maintaining video quality.
Solution Approach 2:
The patent applies different linear prediction models with locally optimized parameters (α and β) to different regions or blocks of video data based on their specific characteristics. This local quality approach allows the system to use simpler models where appropriate while applying more accurate models only where needed, balancing video quality with computational complexity.
Data Source
AI summary
A method for decoding video data, comprising: obtaining a video block from a bitstream; obtaining internal luma sample values of the video block, external luma sample values of an external region of the video block and external chroma sample values of the external region; determining, based on the external luma sample values and the external chroma sample values, a set of weighting coefficients corresponding to a filter shape; predicting, with the filter shape and the set of weighting coefficients, each of internal chroma sample values based on a plurality of corresponding luma sample values, wherein the plurality of corresponding luma sample values comprise: one or more down-sampled luma sample values associated with the filter shape, one or more non-down-sampled luma sample values associated with the filter shape, and one or more non-linear luma sample values; and obtaining predicted video block using the predicted internal chroma sample values.