JVET Intra Mode Coding Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current JVET video coding schemes face inefficiencies in intra mode coding, particularly in coding burdens and bandwidth usage due to the complexity of encoding 67 intra prediction modes, where the last two modes on the MPM list offer no coding advantage and context-based entropy coding increases complexity for MPM modes.
Innovation Solution
The method involves defining unique intra prediction coding modes, instantiating subsets of MPM, selected, and non-selected modes, and encoding these using truncated unary binarization for MPM modes and 4-bit fixed-length codes for selected modes, optimizing the coding process to reduce complexity and bandwidth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If context-based entropy coding is used for MPM modes, then coding precision is improved, but device complexity increases
Solution Approach 1:
The patent segments the intra prediction modes into three distinct sets: MPM modes (most probable modes), selected modes, and non-selected modes. Each set is encoded using a different binarization method tailored to its characteristics. This segmentation allows the use of simpler truncated binary coding for MPM modes while maintaining coding precision through adaptive probability models, thereby reducing device complexity compared to uniform context-based entropy coding across all modes.
Solution Approach 2:
The patent changes the encoding parameter (binarization method) based on the mode category. For MPM modes, truncated binary coding is used with fewer bins, while other modes use different encoding approaches. This parameter change optimizes the balance between coding precision and device complexity by matching the encoding method to the statistical properties of each mode set.
2Adaptability or versatility
If 67 intra prediction modes are fully encoded, then adaptability is improved, but coding burden increases
Solution Approach 1:
The patent divides the 67 intra prediction modes into three segments: MPM modes (typically 6 modes), selected modes (typically 16 modes), and non-selected modes (remaining 45 modes). This segmentation enables differential encoding where each segment uses an optimized binarization method, reducing the overall coding burden while maintaining full adaptability across all 67 modes.
Solution Approach 2:
The patent applies partial action by using truncated binary coding with fewer bins for the MPM modes (the most frequently used modes), while using more comprehensive encoding for less frequent modes. This partial optimization reduces the average coding burden while maintaining full adaptability when needed.
3Adaptability or versatility
If the last two modes on the MPM list are included, then adaptability is improved, but coding efficiency deteriorates
Solution Approach 1:
The patent extracts the last two modes from the MPM list and treats them differently, assigning them to the non-selected modes set. This extraction removes the inefficiency of encoding these rarely used modes with the same resources as the most probable modes, improving overall coding efficiency while retaining adaptability through the comprehensive mode set.
Solution Approach 2:
The patent applies partial optimization by focusing the simplified truncated binary coding approach on the truly most probable modes (excluding the last two), thereby improving coding efficiency for the modes that matter most while still maintaining full adaptability through the complete mode set.
Data Source
AI summary
A method of partitioning a video coding block for JVET, wherein a set of MPMs includes a set of other than 6 intra prediction coding modes and can be encoded using truncated unary binarization, 16 selected intra prediction coding modes can be encoded used 4 bits of fixed length code and remaining non-selected coding modes can be encoded using truncated binary coding and wherein a JVET coding tree unit can be coded as a root node in a quadtree plus binary tree (QTBT) structure that can have a quadtree branching from the root node and binary trees branching from each of the quadtree's leaf nodes using asymmetric binary partitioning to split a coding unit represented by a quadtree leaf node into child nodes, representing the child nodes as leaf nodes in a binary tree branching from the quadtree leaf node.