Geometric Partitioning for Non-Rectangular Image Block Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image encoding/decoding technologies face challenges in efficiently handling high-resolution images, particularly in partitioning and predicting pixel values in non-square and non-rectangular shapes, which affects compression and transmission efficiency.
Innovation Solution
The method involves determining whether geometric partitioning is applicable to a target block based on its size, and performing encoding and decoding on partitioned regions using geometric partitioning, which divides the block into non-square and non-rectangular shapes, allowing for different prediction schemes and merge indices to be applied to these regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional rectangular partitioning is used for high-resolution images, then encoding simplicity is maintained, but partitioning flexibility and prediction accuracy deteriorate
Solution Approach 1:
The target block is segmented into multiple partitioned regions using geometric partitioning lines that can be positioned at arbitrary locations and orientations. This allows the image block to be divided into non-rectangular shapes that better adapt to the actual content structure, improving partitioning flexibility while maintaining manageable encoding complexity through systematic processing of each region.
Solution Approach 2:
The patent introduces asymmetric geometric partitioning where partitioning lines can be positioned at different distances from reference points and at various angles, creating non-uniform, non-rectangular partitioned regions. This asymmetric approach allows better adaptation to image content characteristics while the encoding system manages the increased complexity through standardized geometric descriptions.
2Manufacturing precision
If geometric partitioning into non-square and non-rectangular shapes is applied, then prediction accuracy and compression efficiency are improved, but processing complexity increases
Solution Approach 1:
Different prediction schemes and merge indices are applied to different partitioned regions based on their specific geometric characteristics and content requirements. This local quality approach allows each region to be processed with the most appropriate method, improving overall prediction accuracy while distributing the processing complexity across multiple specialized operations rather than one complex uniform process.
Solution Approach 2:
The patent changes the geometric parameters of partitioning from fixed rectangular grids to variable geometric shapes defined by distance and angle parameters. This allows the partitioning to adapt to local image characteristics, improving prediction accuracy. The processing complexity is managed by encoding these geometric parameters efficiently and using them to guide subsequent processing operations.
3Productivity
If different prediction schemes are applied to multiple partitioned regions, then compression efficiency is improved, but encoding complexity increases
Solution Approach 1:
The encoding system dynamically selects and applies different prediction schemes to different partitioned regions based on their geometric properties and content characteristics. This dynamic approach improves compression efficiency by optimizing the prediction method for each region. The encoding complexity is managed through systematic decision-making processes and efficient representation of the selected schemes.
Solution Approach 2:
The patent employs multiple prediction schemes including intra prediction, inter prediction, and merge mode within a unified geometric partitioning framework. This multi-functionality allows the system to handle diverse image content effectively, improving compression efficiency. The encoding complexity is managed by integrating these different prediction methods into a coherent processing flow that leverages the strengths of each scheme.
Data Source
AI summary
Disclosed herein are a method, an apparatus, and a storage medium for image encoding/decoding using geometric partitioning. A target block may be geometrically partitioned, and multiple partitioned regions may be generated from geometric partitioning. Each partitioned region may have a non-square and a non-rectangular shape. Different prediction schemes may be applied to different partitioned regions. The different prediction schemes for multiple partitioned regions may be signaled as explicit information, and may be derived using information about a coding parameter for the target block or the like.