Matrix-Based Intra Prediction for Selective MIP Signaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images and immersive media such as VR and AR has led to higher transmission and storage costs due to the increased amount of data, necessitating a more efficient image/video compression technology, particularly in matrix-based intra prediction (MIP) processes.
Innovation Solution
Adaptive selection of MIP application based on block type, allowing for MIP processes only where efficiency is high, reducing the number of MIP modes for low-efficiency blocks, and performing MIP without limiting block type conditions to enhance compression efficiency and simplify signaling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If MIP process is applied to all block types, then compression efficiency is improved, but device complexity and signaling overhead increase
Solution Approach 1:
The patent applies MIP process selectively to specific block types (e.g., non-square blocks, blocks with specific width-height ratios) rather than uniformly to all blocks. This local differentiation optimizes compression efficiency for blocks where MIP provides benefit while avoiding unnecessary complexity for blocks where it does not, thereby resolving the contradiction between compression efficiency and device complexity.
2Adaptability or versatility
If MIP process is applied without block type limitations, then adaptability is improved, but loss of information increases due to unnecessary modes
Solution Approach 1:
The patent changes the parameter of block type classification criteria to determine MIP applicability. By defining specific conditions (e.g., width-height ratio thresholds, block shape categories), the patent enables adaptive MIP application that maintains flexibility while eliminating redundant signaling for block types where MIP is not beneficial, thus resolving the contradiction between adaptability and information loss.
Data Source
AI summary
An image decoding method includes obtaining image information including residual information and prediction-related information through a bitstream, deriving transform coefficients based on the residual information, generating residual samples for a current block based on the transform coefficients, deriving an intra prediction mode for the current block based on the prediction-related information, generating prediction samples for the current block based on the intra prediction mode, generating reconstructed samples for the current block based on the prediction samples and the residual samples, wherein the image information includes flag information related to whether an intra prediction mode type for the current block is a matrix-based intra prediction (MIP), wherein the intra prediction mode type for the current block is determined as the MIP based on the flag information, and wherein the prediction samples for the current block are generated based on MIP samples.