Square-to-Rectangular Quantization Matrices for Image Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in existing video encoding methods, such as HEVC and VVC, is the increase in code amount of quantization matrices due to individually defining them for various orthogonal transformation shapes, which affects encoding efficiency.
Innovation Solution
Generating a second quantization matrix from a first quantization matrix of size N×N to a size P×Q, where P and Q are positive integers, allowing quantization of sub-blocks with varying sizes using the second matrix, thereby reducing the need for multiple matrices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If quantization matrices are individually defined for all orthogonal transformation shapes, then the encoding efficiency is improved, but the code amount of quantization matrices unnecessarily increases
Solution Approach 1:
The patent applies universality by enabling a single square quantization matrix to serve multiple functions across different rectangular transformation shapes. The generation unit creates rectangular quantization matrices from a square base matrix, allowing one matrix to adapt to various transformation types (horizontal, vertical, and square) without requiring separate matrices for each shape, thus reducing overall code amount while maintaining encoding efficiency
Solution Approach 2:
The patent utilizes parameter changes by modifying the dimensions of the quantization matrix from a fixed square shape to variable rectangular shapes. The generation unit transforms a square quantization matrix into rectangular matrices with different aspect ratios based on the transformation shape, allowing the same base matrix to adapt to different parameter requirements without increasing code amount
2Quantity of substance
If a single quantization matrix is used for all transformation shapes, then the code amount is reduced, but the encoding efficiency deteriorates due to lack of shape-specific optimization
Solution Approach 1:
The patent applies segmentation by dividing the quantization process into base matrix definition and shape-specific generation. The square quantization matrix serves as a base that is then segmented and transformed into different rectangular configurations according to the specific transformation shape requirements, allowing both code reduction and shape-specific optimization
Solution Approach 2:
The patent implements dynamics by making the quantization matrix adaptable and flexible rather than static. The generation unit dynamically creates appropriate rectangular quantization matrices from the square base matrix based on the actual transformation shape being used, allowing the system to optimize for each specific case while maintaining a compact representation
Data Source
AI summary
An image encoding apparatus for encoding an image comprises a generation unit configured to generate, from a first quantization matrix having a size of N×N (N is a positive integer), a second quantization matrix having a size of P×Q (P and Q are positive integers which satisfy P<N<Q, and the size of P×Q indicates a size of P in a horizontal direction and a size of Q in a vertical direction), and a quantization unit configured to quantize transformation coefficients in a sub-block having a size corresponding to the size of P×Q using the second quantization matrix.