Video Codec Sample-Level Weights for Brightness-Adaptive Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding technologies face inaccuracies in inter prediction due to fixed weight values used for brightness changes, leading to reduced compression performance.
Innovation Solution
Determine sample-level weight parameters through type indication and geometrical mode parameters during encoding and decoding to adapt to changes in brightness, improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If fixed weight values are used for inter prediction in a unit of slice or coding unit, then the prediction process is simple and fast, but the accuracy of inter prediction values is low, reducing compression performance
Solution Approach 1:
The patent applies local quality by transitioning from fixed slice-level or coding unit-level weight values to sample-level weight parameters that are specific to each sample's position and characteristics. The weight parameter determination is based on sample position information and geometrical mode parameters, allowing different weights for different samples within the same block to accurately model local brightness changes and improve prediction accuracy.
Solution Approach 2:
The patent implements dynamics by making weight parameters adaptive rather than fixed. The weight parameter determination process uses sample position information and geometrical mode parameters to dynamically calculate appropriate weights for each sample, allowing the prediction system to adapt to varying brightness changes in different regions of the video content.
2Measurement precision
If sample-level weight parameters are determined using type indication and geometrical mode parameters, then prediction accuracy is improved, but the encoding and decoding process becomes more complex
Solution Approach 1:
The patent applies preliminary action by pre-defining the relationship between geometrical mode parameters and weight parameters. The geometrical mode parameters are determined during motion estimation, and the corresponding weight parameters are pre-calculated based on these geometrical modes, allowing the actual prediction process to simply retrieve and apply the pre-determined weights without complex real-time calculations.
Solution Approach 2:
The patent uses geometrical mode parameters as an intermediary between the motion estimation process and the weight parameter determination. The geometrical mode parameters serve as a bridge that connects the spatial transformation information to the weight parameter selection, simplifying the overall process by using an intermediate representation that directly maps to weight parameters.
3Productivity
If fixed weight values are used for inter prediction, then the encoding and decoding process is efficient, but compression performance is reduced due to low prediction accuracy
Solution Approach 1:
The patent applies parameter changes by transitioning from fixed weight values to variable weight parameters that are determined based on sample position information and geometrical mode parameters. This parameter change allows the system to adapt weights to specific local conditions, improving prediction accuracy and consequently compression performance while maintaining encoding efficiency through systematic parameter determination.
Data Source
AI summary
The present application provides an encoding and decoding method, a codec, a bitstream, and a storage medium. The codec determines at least one piece of motion vector information of a current block, and a type indication parameter and/or a geometric mode parameter (101), determines at least one reference predicted value of the current block according to the at least one piece of motion vector information, and determines at least one weight parameter of the current block according to the type indication parameter and/or the geometric mode parameter (102), and determines a predicted value of the current block on the basis of the at least one reference predicted value and the at least one weight parameter (103).