Image Intra Prediction Using Split MPM Candidate Groups
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compression technologies face challenges in accurately and efficiently determining intra prediction modes, particularly for high-resolution and high-quality images, which affect the efficiency of image encoding and decoding processes.
Innovation Solution
An image encoding/decoding method and apparatus that determines a reference region for intra prediction of a current block, divides the MPM candidate group into two groups, and derives the intra prediction mode based on these groups, using predefined pixel lines and flags to select the appropriate mode for luma and chroma blocks, with adaptive block division and inter-component reference techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If intra prediction mode is derived using a single unified method, then the encoding process is simple, but the prediction accuracy for different block types is insufficient
Solution Approach 1:
The patent segments the MPM candidate group into two distinct groups: a first group containing only non-directional modes (planar and DC modes) and a second group containing directional modes. This segmentation allows the encoder to selectively apply different prediction strategies based on block characteristics, improving prediction accuracy for different block types while maintaining manageable encoding complexity through structured organization.
Solution Approach 2:
The patent introduces dynamic selection mechanisms where the encoder can choose between using the first group, second group, or both groups together based on block characteristics. The use of flags to indicate which group to use and the ability to derive modes by adding or subtracting N values from neighboring block modes creates a dynamic system that adapts to different prediction scenarios, enhancing accuracy without requiring a completely separate method for each case.
2Measurement precision
If more MPM candidate modes are used, then the prediction accuracy improves, but the bitstream overhead increases
Solution Approach 1:
By dividing the MPM candidate group into two separate groups with distinct characteristics (non-directional vs. directional modes), the patent enables more efficient encoding. The encoder can signal which group to use with minimal overhead and select the most appropriate mode within that group, reducing the number of bits needed to represent the prediction mode while maintaining comprehensive coverage of possible prediction scenarios.
Solution Approach 2:
The patent introduces parameters such as the first flag indicating which group to use, and N values (1, 2, or 3) for deriving modes by adding or subtracting from neighboring block modes. These parameter changes allow the system to efficiently represent a larger number of possible prediction modes through compact signaling, reducing bitstream overhead while maintaining the ability to derive accurate prediction modes for diverse block types.
3Productivity
If adaptive block division is implemented, then the prediction efficiency improves, but the processing complexity increases
Solution Approach 1:
The patent applies segmentation to the block division process by dividing the current block into multiple sub-blocks and performing intra prediction separately for each sub-block. This allows the encoder to adapt to local variations within the block, improving prediction efficiency for blocks with different characteristics in different regions. The structured approach to sub-block processing manages complexity through systematic organization of the division and prediction operations.
Data Source
AI summary
A method and a device for encoding/decoding an image according to the present invention may determine a reference region for intra prediction of a current block, derive an intra prediction mode of the current block on the basis of a predetermined MPM candidate group, and perform intra prediction on the current block on the basis of the reference region and the intra prediction mode.