GPM Split Mode Reordering for Intra-Inter Video Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards like VVC struggle with achieving superior coding efficiency, and there is a need for further improvements in geometric partitioning (GPM) mode to enhance compression performance.
Innovation Solution
The proposed methods involve disabling template matching (TM) based GPM split modes reordering for CUs predicted with intra prediction, and enabling it for CUs predicted with inter prediction, while utilizing inter and intra prediction samples to reorder and select optimal GPM split modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If template matching based GPM split modes reordering is applied to all CUs, then coding efficiency is improved for inter prediction, but computational complexity increases and performance degrades for intra prediction
Solution Approach 1:
The patent applies different processing strategies to different regions (prediction types) within the coding space. Template matching based reordering is selectively applied only to inter prediction parts while intra prediction parts use conventional reordering, optimizing performance for each local region's characteristics
2Measurement precision
If template matching based GPM split modes reordering is applied, then prediction accuracy is improved, but processing time increases
Solution Approach 1:
The patent applies template matching based reordering selectively only to inter prediction parts rather than universally to all prediction types. This partial application achieves the accuracy benefits where needed while avoiding the processing time penalty in intra prediction scenarios
3Speed
If GPM split modes reordering is disabled for intra prediction, then processing speed is maintained, but coding efficiency is suboptimal compared to inter prediction
Solution Approach 1:
The patent recognizes that intra and inter prediction parts have different characteristics and optimizes the reordering approach locally for each type, using conventional methods for intra prediction to maintain speed while applying template matching methods to inter prediction to maximize coding efficiency
Data Source
AI summary
Methods for video decoding and encoding, apparatuses and non-transitory storage media are provided. In one decoding method, the decoder obtains a first geometric partition mode (GPM) part and a second GPM part for a coding unit (CU). The decoder disables template matching (TM) based GPM split modes reordering for multiple GPM split modes that are applied to the CU in response to determining that the first GPM part or the second GPM part is predicted with intra prediction. Furthermore, the decoder enables the TM based GPM split modes reordering for the multiple GPM split modes in response to determining that both the first GPM part and the second GPM part are predicted with inter prediction.