Matrix-Based Intra-Prediction Mode Indexing for Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current matrix-based intra-prediction (MIP) methods face challenges in memory requirements for storing extra MIP-matrices and inefficiencies in signaling costs, especially when dealing with transposed modes and blocks with aspect ratios greater than four.
Innovation Solution
The proposed solution involves using a list of matrix-based intra-prediction modes where all MIP modes are applicable in a transposed way, reducing the number of modes and eliminating the need for additional syntax elements to indicate transposition. This is achieved by grouping MIP-modes into pairs with the same prediction matrix, allowing the mode index to determine both the prediction mode and its transposed condition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If extra MIP-matrices are stored for non-transposed modes, then prediction accuracy is improved, but memory requirements increase
Solution Approach 1:
The patent merges transposed and non-transposed MIP modes by storing only one set of prediction matrices. The decoder determines whether to apply transposition based on the mode index, eliminating the need to store separate matrices for both transposed and non-transposed variants. This reduces memory requirements while maintaining prediction accuracy through intelligent reuse of the same matrices with different orientation applications.
Solution Approach 2:
The stored prediction matrices are made universal by designing them to serve both transposed and non-transposed modes. The same set of matrices can be applied in different orientations depending on the block shape and mode selection, making the storage system multi-functional and eliminating redundant matrix storage for different block aspect ratios.
2Measurement precision
If additional syntax elements are used to indicate transposed modes, then mode precision is improved, but signaling costs increase
Solution Approach 1:
The patent merges the mode index and transposition indication into a single syntax element. The mode index simultaneously identifies the prediction mode and implicitly indicates whether transposition should be applied, based on the relationship between the mode index value and the block aspect ratio. This eliminates the need for separate transposition flags or additional syntax elements.
Solution Approach 2:
Instead of explicitly signaling transposition with additional bits, the patent inverts the approach by using the mode index to implicitly indicate transposition requirements. The decoder infers whether transposition is needed based on the selected mode and block characteristics, rather than receiving direct transposition commands in the bitstream.
3Device complexity
If MIP modes are restricted for blocks with aspect ratio greater than four, then decoding complexity is reduced, but adaptability decreases
Solution Approach 1:
The patent makes the MIP mode system universal by enabling the same prediction matrices and decoding logic to handle both standard blocks (aspect ratio ≤ 4) and extended blocks (aspect ratio > 4). The mode selection and transposition logic automatically adapts to different block shapes without requiring separate decoding paths or additional complexity.
Solution Approach 2:
The patent introduces dynamic adaptability where the application of MIP modes and transposition is determined on-the-fly based on the block aspect ratio and mode index. Rather than having fixed restrictions, the system dynamically selects appropriate matrix applications and transposition operations based on the specific block characteristics, enabling versatile handling of various block shapes.
Data Source
AI summary
A method for decoding a predetermined block of a picture using intra-prediction by reading, for each of predetermined intra-prediction blocks, from a data stream, a mode index identifying a matrix-based intra-prediction mode is disclosed. Samples of each predetermined intra-prediction block are predicted by computing a matrix-vector product between an input vector derived from reference samples and a prediction matrix associated with the identified matrix-based intra-prediction mode (k) and associating components of an output vector obtained by the matrix-vector product onto sample positions of the predetermined block. Further predetermined intra-predicted blocks of the picture are predicted to obtain a prediction signal. A transformation flag is decoded from the data stream using context adaptive binary arithmetic coding, and a prediction residual is decoded and re-transformed using a reverse transformation based on the transformation flag to obtain a prediction residual signal. The prediction residual signal is used to correct the prediction signal.