Bi-directional CU Weighting for Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face inefficiencies in motion vector prediction and coding due to redundancy issues and parsing errors in the candidate list construction, leading to decoding failures and visual artifacts.
Innovation Solution
The implementation of bi-directional CU weighting (BCW) allows for multiple pairs of weights to be used, enabling more efficient motion compensation and prediction by generating a Merge candidate list with pre-defined weights for bi-directional predictors, and inheriting weights from target Merge candidates, which simplifies the encoding and decoding process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional motion vector prediction is used, then decoding complexity is reduced, but coding efficiency deteriorates due to redundancy and parsing errors
Solution Approach 1:
The patent applies preliminary action by pre-defining weight pairs and storing them in a weight table before the actual coding process. The syntax element signaling mechanism is prepared in advance, allowing the decoder to directly retrieve and apply weights without complex real-time calculations, thus improving coding efficiency while maintaining manageable decoding complexity
Solution Approach 2:
The patent changes the parameter representation by introducing a compact syntax element that indexes into a pre-defined weight table. Instead of transmitting full weight values, the system transmits index parameters that reference stored weight pairs, reducing redundancy and improving coding efficiency without significantly increasing decoding complexity
2Adaptability or versatility
If multiple weight pairs are supported, then coding flexibility is improved, but device complexity increases due to weight management
Solution Approach 1:
The patent implements universality by creating a weight table that serves multiple functions: it stores pre-defined weight pairs for different prediction scenarios, provides a reference mechanism for both encoder and decoder, and supports various bi-directional prediction modes through a unified syntax element. This multi-functional approach enables coding flexibility while avoiding the need for separate weight management mechanisms for each scenario
Solution Approach 2:
The patent uses copying by replicating the same weight table and syntax element interpretation logic in both encoder and decoder. Instead of complex weight management, the system copies pre-defined weight pairs into a standardized table structure that both sides reference, ensuring consistency while simplifying implementation
3Loss of energy
If Merge mode is used, then bandwidth consumption is reduced, but prediction accuracy deteriorates due to candidate list redundancy
Solution Approach 1:
The patent applies local quality by associating specific weight pairs with specific Merge candidates in the candidate list. Instead of using uniform weights for all candidates, the system assigns locally optimized weights to individual candidates based on their characteristics, improving prediction accuracy while maintaining the low bandwidth advantage of Merge mode
Solution Approach 2:
The patent uses preliminary action by pre-ordering and pre-weighting the Merge candidate list before transmission. The encoder prepares the candidate list with associated weights in advance, allowing the decoder to directly use these pre-optimized candidates without requiring additional signaling or complex selection processes, thus maintaining low bandwidth consumption while improving prediction accuracy
Data Source
AI summary
A method and apparatus of encoding or decoding for a coding system, where bi-directional CU (coding unit) weighting (BCW) is enabled for the coding system to allow using multiple pairs of weights, are disclosed. According to this method, a Merge candidate list is generated for the current block, where current bi-directional CU weights for a bi-directional predictor of the current block associated with a target Merge candidate in the Merge candidate list are set to a pre-define pair of weights when the target Merge candidate corresponds to a selected Merge candidate. In another method, current bi-directional CU weights for a bi-directional predictor associated with a target Merge candidate in the Merge candidate list are inherited from the target Merge candidate when the target Merge candidate corresponds to a selected Merge candidate.