Intra Prediction Mode Selection Using Texture and Statistics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image coding systems face challenges in determining the most probable intra-prediction mode for efficient compression, as they fail to effectively utilize spatial correlation between adjacent pixels, leading to suboptimal compression efficiency.
Innovation Solution
A system and method that determine a final most probable mode (MPM) list for intra-prediction by considering first intra-prediction modes of adjacent blocks, block dimensions, statistical information of encoded blocks, and texture features, allowing for the selection of the target intra-prediction mode based on this list to encode the target block.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional intra-prediction methods are used, then the encoding process is simple, but compression efficiency is suboptimal
Solution Approach 1:
The patent applies preliminary action by pre-determining and storing MPM lists for different block sizes and configurations before actual encoding. The system pre-calculates statistical information and texture features for various block types, so that during encoding, the MPM list can be directly retrieved or slightly adjusted rather than computed in real-time, significantly reducing encoding complexity while maintaining high compression efficiency
Solution Approach 2:
The patent utilizes parameter changes by adapting the MPM list determination method based on different block dimensions, positions, and image regions. The system changes parameters such as block size, reference block selection, and statistical weighting factors according to the specific encoding context, allowing the system to optimize compression efficiency for different scenarios without using a single fixed complex method
2Productivity
If spatial correlation between adjacent pixels is not utilized, then the processing is faster, but compression efficiency deteriorates
Solution Approach 1:
The patent introduces an intermediary mechanism that processes spatial correlation information in advance and stores it in a structured format. The MPM list acts as an intermediary that pre-incorporates spatial correlation knowledge, allowing the encoding process to benefit from spatial correlation without performing complex real-time calculations, thus balancing compression efficiency with processing speed
Solution Approach 2:
The system performs preliminary analysis of spatial correlation patterns and pre-determines MPM lists that capture these relationships. By preparing this information beforehand, the actual encoding process can quickly select appropriate prediction modes without performing extensive spatial correlation analysis in real-time, maintaining both efficiency and speed
3Measurement precision
If a comprehensive MPM list determination method is used, then prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing MPM lists with high prediction accuracy for various block configurations. The comprehensive analysis of spatial correlation, statistical information, and texture features is performed in advance, so that during actual encoding, the system only needs to retrieve and possibly adjust pre-computed MPM lists, achieving high prediction accuracy without the computational complexity of real-time comprehensive analysis
Solution Approach 2:
The patent segments the complex MPM list determination process into separate stages: pre-computation of statistical information and texture features for different block types, storage of MPM lists, and retrieval/adaptation during encoding. This segmentation allows the system to achieve comprehensive prediction accuracy through pre-computed data while keeping the actual encoding process computationally simple
Data Source
AI summary
A method for intra prediction may include obtaining a target block to be encoded in a target image. The method may also include determining a final most probably mode (MPM) list of the target block based on at least one of (a) first intra-prediction modes configured to encode two or more adjacent blocks of the target block, (b) a width and a height of the target block, (c) statistical information of second intra-prediction modes configured to encode a plurality of first encoded blocks associated with the target block, or (d) texture features of a plurality of second encoded blocks associated with the target block. The method may also include determining, based on the final MPM list, a target intra-prediction mode corresponding to the target block. The method may also include encoding, based on the target intra-prediction mode, the target block.