Wide-Angle Intra Prediction for Rectangular Video Sub-Partitions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding technologies face inefficiencies in intra prediction for rectangular sub-partitions due to the unavailability of reference samples and the use of fixed prediction directions that do not adapt well to block shapes, leading to sub-optimal compression efficiency.
Innovation Solution
Adapt the intra prediction direction for rectangular sub-partitions based on their dimensions, using wide-angle modes when necessary, and substitute missing reference samples from the parent CU to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If fixed prediction directions are used for rectangular sub-partitions, then the encoding process is simple, but compression efficiency is sub-optimal
Solution Approach 1:
The patent implements dynamic prediction direction selection for rectangular sub-partitions based on their aspect ratios. Instead of using fixed prediction directions, the system adapts the prediction mode (horizontal, vertical, or diagonal) according to the specific dimensions of each sub-partition, allowing the prediction process to dynamically adjust to block shapes while maintaining reasonable complexity through predefined adaptation rules.
Solution Approach 2:
The patent changes the prediction direction parameters based on the aspect ratio of rectangular sub-partitions. By modifying the prediction angle and direction parameters according to block geometry, the system optimizes prediction accuracy for different block shapes, improving compression efficiency without requiring complete redesign of the prediction framework.
2Measurement precision
If reference samples are not available for rectangular sub-partitions, then the encoding process is simple, but prediction accuracy is reduced
Solution Approach 1:
The patent introduces parent coding unit (CU) reference samples as an intermediary source to substitute missing reference samples for rectangular sub-partitions. When direct reference samples from above or left of the sub-partition are unavailable, the system uses reference samples from the parent CU, effectively bridging the gap and enabling accurate prediction without requiring complex reference sample generation mechanisms.
3Productivity
If conventional intra prediction modes are used for all blocks, then the encoding process is uniform and simple, but performance is sub-optimal for non-square blocks
Solution Approach 1:
The patent applies local quality adaptation by using different prediction modes for different regions (sub-partitions) based on their local geometric characteristics. Rectangular sub-partitions with specific aspect ratios receive specialized prediction treatments (horizontal, vertical, or diagonal modes) tailored to their local shape, while other blocks use conventional modes, achieving optimal performance for each local region.
Solution Approach 2:
The patent modifies prediction parameters (direction, angle, reference sample selection) based on block shape characteristics. For rectangular sub-partitions with aspect ratios indicating horizontal or vertical orientation, the system changes the prediction parameters accordingly, enabling adaptive performance optimization across different block geometries while maintaining a unified encoding framework.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and apparatus for performing prediction for encoding or decoding uses intra prediction with sub-partitions. The sub-partitions are oriented either horizontally or vertically and can use wide-angle modes different than that of the video block they originate from. Reference samples for a sub-partition are those of the video block when reference samples for the sub-partition are not available, such as due to direction. In an embodiment, when a sub-partition is square, conventional intra prediction directions are used. Reference samples can be used from a block above and right of the video block or left and below the video block using a mapping along a prediction direction, vertically, or horizontally.