Bi-predictive Weighting Parameter Coding via Differential Zero Tree
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, such as MPEG-1/2 and H.264, face inefficiencies in coding motion and prediction weighting parameters, particularly in handling local brightness variations and bi-predictive motion compensation, which limits their ability to improve coding efficiency in scenes with temporal brightness variations.
Innovation Solution
A method and apparatus for efficiently coding weighting parameters in bi-predicted partitions by transforming and signaling these parameters using a zero tree coding structure, allowing for variable length coding and differential coding to reduce overhead, while considering both global and local weighting parameters and their dynamic range.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If weighting parameters are transmitted explicitly for each reference in bi-predictive motion compensation, then coding efficiency is improved by handling local brightness variations, but overhead bits increase significantly
Solution Approach 1:
The patent extracts and transmits only the differential values of weighting parameters rather than the complete parameter sets. By taking out only the essential variation information from the reference picture weighting parameters and transmitting these differences, the overhead is significantly reduced while maintaining the ability to represent local brightness variations accurately.
Solution Approach 2:
The patent performs preliminary encoding of weighting parameters at the picture level, establishing reference weighting parameters before block-level processing. This preliminary action allows subsequent blocks to use differential coding relative to these pre-established references, reducing the total overhead required for representing all weighting parameters throughout the picture.
2Adaptability or versatility
If multiple weighting parameters are used for each reference in bi-prediction, then handling of local brightness variations is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by allowing different weighting parameters to be used for different spatial regions (blocks) within a picture. Each block can have its own differential weighting parameters that are adapted to local brightness characteristics, while still using the same reference picture. This enables localized adaptation without requiring completely independent parameter sets for each region.
Solution Approach 2:
The patent changes the parameter representation from absolute weighting values to differential values relative to reference pictures. This parameter transformation simplifies the coding process by exploiting the correlation between neighboring blocks' weighting parameters, reducing the complexity of transmitting and processing multiple weighting parameters while maintaining adaptability to local variations.
3Quantity of substance
If global weighting parameters are used for every frame, then overhead is reduced, but ability to handle local brightness variations is insufficient
Solution Approach 1:
The patent segments the picture into multiple blocks and allows each block to have its own differential weighting parameters. This segmentation enables local adaptation to brightness variations while using a common reference framework. The picture is divided into manageable units that can independently adapt to local conditions without requiring separate global parameters for each block.
Solution Approach 2:
The patent introduces reference picture weighting parameters as intermediaries between global picture-level weighting and local block-level weighting. These reference parameters serve as mediators that capture picture-wide trends while allowing blocks to add local variations through differential coding. This intermediary layer enables both global efficiency and local adaptability.
Data Source
Figure 1A~1B
Figure 2~3
Figure 4
AI summary
A method and apparatus is disclosed herein for encoding and/or decoding are described. In one embodiment, the encoding method comprises generating weighting parameters for multi-hypothesis partitions, transforming the weighting parameters and coding transformed weighting parameters.