Multi-Level MPM Sets for Intra Prediction Mode Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current High-Efficiency Video Coding (HEVC) methods for intra prediction mode coding, particularly for luma and chroma components, face inefficiencies due to the use of fixed length binarization, which is not suitable for non-uniform probability distributions, leading to poor coding efficiency.
Innovation Solution
The implementation of multi-level Most Probable Mode (MPM) sets for intra prediction mode encoding and decoding, where intra prediction modes are configured into multiple levels, allowing for adaptive coding of remaining modes using n-bit fixed length binarization and variable-length codewords, improving coding efficiency by skipping redundant symbols and utilizing predefined or adaptive tables.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If fixed length binarization is used for intra prediction mode coding, then the coding process is simple, but the coding efficiency is poor due to non-uniform probability distributions
Solution Approach 1:
The patent segments the intra prediction mode coding process into multiple levels. The first level uses a simplified fixed-length binarization for the most probable modes, while subsequent levels handle remaining modes with more sophisticated coding. This segmentation allows the system to achieve good coding efficiency without requiring complex coding for all modes, thus resolving the contradiction between simplicity and efficiency.
Solution Approach 2:
The patent introduces dynamic adaptation by using different binarization strategies based on the probability distribution of prediction modes. The coding scheme dynamically switches between fixed-length and variable-length coding depending on whether the mode is in the most probable set or remaining modes, optimizing efficiency while maintaining reasonable complexity.
2Measurement precision
If multiple intra prediction modes are used to exploit spatial features, then the spatial prediction accuracy is improved, but the side information in the bitstream increases
Solution Approach 1:
The patent segments the set of 35 intra prediction modes into two groups: most probable modes (MPM) and remaining modes. By coding MPM modes with fewer bits and using adaptive coding for remaining modes, the patent reduces the overall side information volume while maintaining the ability to represent all 35 modes, thus resolving the contradiction between prediction accuracy and bitstream size.
Solution Approach 2:
The patent changes the coding parameters (number of bits) based on the probability distribution of different prediction modes. High-probability modes are assigned shorter codewords, while low-probability modes use longer codewords or are coded differently, optimizing the balance between representation accuracy and bitstream efficiency.
3Reliability
If luma intra prediction mode information is incorporated in the bitstream for each PU, then the decoder can correctly reconstruct the image, but the bitstream size becomes substantial
Solution Approach 1:
The patent segments luma intra prediction mode coding into a two-level process: first coding the most probable mode index with fewer bits, then conditionally coding remaining mode information only when necessary. This segmentation ensures reliable reconstruction while reducing the average bitstream size compared to coding all modes with the same number of bits.
Solution Approach 2:
The patent applies partial coding by only transmitting full mode information when the actual mode falls into the remaining modes category. For most cases where the mode is in the MPM set, a simplified coding path is used, reducing the overall bitstream volume while maintaining complete reconstruction capability when needed.
Data Source
AI summary
A method for intra prediction mode decoding for a current block based on multi-level most probable mode (MPM) sets. The method includes determining one or more MPM candidates for a first-level MPM set, and determining one or more most probable remaining mode (MPRM) candidates for a second-level MPM set. Wherein the one or more MPRM candidates are identified based on a mode of a neighboring block of the current block, and the one or more MPRM candidates are different from any MPM candidate in the first-level MPM set. The method also includes decoding syntax information related to a current intra prediction mode corresponding to the current block according to the multi-level MPM sets using a decoder, and decoding the current block based on the determined current intra prediction mode.