Geometric Partitioning of Image Blocks for Adaptive Intra-Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image encoding/decoding technologies face challenges in efficiently handling high-resolution and high-definition images, particularly in applying intra-prediction to improve compression efficiency.
Innovation Solution
The method involves geometric partitioning of a target block into subblocks, applying identical intra-prediction modes to these subblocks, and constructing independent Most Probable Mode lists for each subblock, utilizing geometric partitioning mode information for prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If geometric partitioning is applied to target blocks for intra-prediction, then compression efficiency is improved, but device complexity increases
Solution Approach 1:
The patent applies geometric partitioning to divide target blocks into multiple subblocks based on triangular or quadrangular partitioning patterns. This segmentation allows different intra-prediction modes to be applied to different subblocks, improving compression efficiency by adapting to local image characteristics while maintaining manageable complexity through systematic partitioning rules
Solution Approach 2:
The patent enables different intra-prediction modes to be applied to different subblocks within the same target block. This local quality approach allows the encoding system to adapt prediction parameters to local image characteristics, improving compression efficiency without requiring complete redesign of the entire encoding process
2Measurement precision
If multiple subblocks are created through geometric partitioning, then prediction accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments target blocks into multiple subblocks using geometric partitioning patterns (triangular or quadrangular). This segmentation enables more accurate prediction by allowing different intra-prediction modes for different subblocks, while the systematic nature of the partitioning keeps processing time manageable
Solution Approach 2:
The patent applies geometric partitioning selectively to target blocks where it provides benefit, rather than uniformly applying it to all blocks. This partial action approach improves prediction accuracy for complex regions while avoiding unnecessary processing time consumption in simpler regions
3Adaptability or versatility
If independent MPM lists are constructed for each subblock, then encoding flexibility is improved, but data complexity increases
Solution Approach 1:
The patent divides the MPM list construction process into independent segments for each subblock. This allows each subblock to have its own MPM list tailored to local characteristics, improving encoding flexibility while the structured approach to list construction manages bitstream complexity
Solution Approach 2:
The patent constructs independent MPM lists for each subblock, allowing local adaptation of prediction modes. This local quality approach improves encoding flexibility by enabling subblock-specific optimization while controlling data complexity through systematic list construction rules
Data Source
AI summary
Disclosed herein are a method, an apparatus and a storage medium for image encoding/decoding using geometric partitioning. Through geometric partitioning, a target block is partitioned into a first partitioned region and a second partitioned region. Intra-prediction is used for at least one of the first partitioned region and the second partitioned region. An intra-prediction mode of intra-prediction used for the partitioned region may be limited by a partition boundary or the like of a geometric partitioning mode. Through prediction for the first partitioned region and prediction for the second partitioned region, a prediction block for the target block is generated.