VVC Encoding With Universal Quantization Matrices for Rectangular Transforms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in VVC encoding is the increased 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 accommodate rectangular orthogonal transformations, allowing quantization of sub-blocks of size P×Q, where P and Q are positive integers with specific directional relationships, 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 creating a single square quantization matrix that serves multiple functions for different orthogonal transformation shapes. Instead of defining separate quantization matrices for each shape (horizontal, vertical, diagonal), the invention uses one universal square matrix that can be adaptively applied to various rectangular and non-rectangular transformation outcomes, thereby reducing the total code amount while maintaining encoding efficiency.
Solution Approach 2:
The patent employs parameter changes by dynamically selecting and applying different portions or scaling factors of the square quantization matrix based on the specific orthogonal transformation shape being used. The quantization parameters are adjusted according to the transformation type, allowing the same base matrix to effectively handle multiple shapes without requiring separate predefined matrices for each.
2Reliability
If quantization matrices are individually defined for all orthogonal transformation shapes, then the image quality is maintained, but the device complexity increases
Solution Approach 1:
The patent reduces device complexity by implementing a universal quantization matrix that handles multiple orthogonal transformation shapes. This eliminates the need for complex matrix management systems that would otherwise be required to store, select, and manage multiple shape-specific quantization matrices, while still maintaining image quality through adaptive parameter application.
Solution Approach 2:
The patent merges the functionality of multiple shape-specific quantization matrices into a single square quantization matrix. By combining what would have been separate matrices into one unified structure, the system reduces complexity in matrix storage, retrieval, and management while preserving the ability to maintain image quality across different transformation types.
3Reliability
If multiple quantization matrices are used for different sub-block shapes, then the rate-distortion performance is optimized, but the data amount increases
Solution Approach 1:
The patent maintains optimized rate-distortion performance by dynamically changing quantization parameters based on the sub-block shape being processed. Instead of storing multiple complete matrices, the system uses a single square matrix with parameters that are adaptively adjusted according to the specific orthogonal transformation shape, thereby reducing data amount while preserving the optimization benefits.
Solution Approach 2:
The patent applies local quality by allowing different regions or portions of the square quantization matrix to be applied to different sub-block shapes. Each sub-block receives appropriately tailored quantization parameters from the universal matrix based on its specific shape characteristics, maintaining optimized rate-distortion performance for each local region without requiring separate global matrices.
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.