Gradient Linear Model for Cross-Component Video Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in achieving superior coding efficiency, particularly in reducing cross-component redundancy, which affects the quality and bit-rate of compressed video data.
Innovation Solution
The proposed method employs a gradient linear model (GLM) for cross-component prediction in video coding. This involves obtaining indications from a bitstream related to the GLM, using it to calculate filtered values based on intensity differences among luma samples, and then encoding or decoding video data accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional prediction methods are used for video coding, then the coding process is simple, but cross-component redundancy is high and coding efficiency is limited
Solution Approach 1:
The patent applies parameter changes by introducing a gradient linear model that transforms the prediction approach from traditional methods to a model-based approach using gradient information and linear combinations of reference samples. This changes the parameters of the prediction system to achieve better coding efficiency while managing complexity through structured parameterization.
Solution Approach 2:
The patent uses an intermediary approach by introducing a gradient linear model as a mediator between the reference samples and the current block prediction. This intermediary model processes the relationship between chroma and luma components through gradient calculations and linear combinations, enabling more efficient cross-component prediction while maintaining systematic complexity management.
2Reliability
If cross-component prediction is enhanced to reduce redundancy, then video quality improves at lower bit-rates, but the complexity of the encoding/decoding process increases
Solution Approach 1:
The patent applies segmentation by dividing the prediction process into distinct stages: obtaining reference samples, calculating gradient information, forming linear combinations, and generating predictions. This segmentation of the complex prediction process into manageable steps enables enhanced cross-component prediction while maintaining systematic complexity management through structured processing stages.
Solution Approach 2:
The patent uses parameter changes by transforming the prediction model to use gradient information and linear combination parameters rather than traditional fixed prediction coefficients. This parameter transformation enables adaptive cross-component prediction that improves video quality while managing complexity through parameterized rather than fixed complex operations.
Data Source
AI summary
The present disclosure provides a method for decoding video data, comprising: obtaining a bitstream; obtaining an indication from the bitstream indicative of information related to a gradient linear model (GLM), wherein the GLM is used to obtain one or more filtered values based on intensity differences among luma samples; and decoding the video data based on the information related to the GLM.