Unified MIP Mode Mapping Table for Video Coding Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing methods for mapping between Matrix-based Intra Prediction (MIP) modes and non-MIP modes in video coding are complex, leading to increased computational complexity and memory usage in both encoding and decoding processes.
Innovation Solution
A simplified mapping method that determines an intra prediction mode for a current block and maps MIP modes to pre-specified non-MIP modes or vice versa, reducing the need for multiple table queries and memory storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple mapping tables are used for mapping between MIP mode and non-MIP mode, then mapping accuracy is improved, but device complexity and memory usage increase
Solution Approach 1:
The patent merges multiple mapping tables (65-33 mapping table and conventional-MIP mapping table) into a single unified mapping table that directly maps between MIP modes and non-MIP modes. This consolidation eliminates the need for sequential table lookups, reducing computational complexity while maintaining mapping accuracy through a streamlined single-table approach.
Solution Approach 2:
The patent extracts and eliminates redundant intermediate mapping steps from the original multi-table process. By removing the need for sequential mapping through multiple tables, the invention simplifies the overall mapping process while preserving the essential functionality of converting between MIP and non-MIP modes accurately.
2Adaptability or versatility
If multiple mapping tables are stored in memory for mode conversion, then mapping completeness is improved, but memory space consumption increases
Solution Approach 1:
The patent combines multiple separate mapping tables into one integrated mapping table that contains all necessary mappings between MIP and non-MIP modes. This single table approach maintains complete mapping coverage while significantly reducing memory consumption by eliminating redundant data storage across multiple tables.
3Manufacturing precision
If complex mapping processes are implemented for MIP and non-MIP mode conversion, then encoding accuracy is improved, but encoding rate and decoding rate decrease
Solution Approach 1:
The patent removes unnecessary intermediate mapping steps from the encoding and decoding process. By implementing a direct single-table mapping approach, the invention reduces computational overhead and processing time while maintaining encoding accuracy, thereby improving both encoding rate and decoding rate.
Solution Approach 2:
The patent enables the system to skip intermediate mapping stages by using a direct mapping table. This allows the encoder and decoder to perform mode conversion more quickly by rushing through the mapping process in a single step rather than sequentially processing through multiple tables, thus improving processing speed.
Data Source
AI summary
Disclosed in the implementations of the present disclosure are a mapping method, an encoder, a decoder, and a computer storage medium. The method may include: determining an intra prediction mode used at the time of encoding or decoding a current block; mapping, if the intra prediction mode is a MIP mode, the MIP mode to a first non-MIP mode; and mapping, if the intra prediction mode is a non-MIP mode, the non-MIP mode to a second MIP mode.