Image Decoding Quantization Matrix for VVC Frequency Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the context of Versatile Video Coding (VVC), the use of zeroing out techniques for orthogonal transform coefficients limits the ability to perform quantization control based on frequency components, thereby hindering improvements in subjective image quality.
Innovation Solution
An image decoding apparatus is designed to enable quantization processing using a quantization matrix corresponding to the zeroing out technique, allowing for effective quantization control of orthogonal transform coefficients even after zeroing out has been applied.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If zeroing out technique is used to reduce code amount, then compression efficiency is improved, but quantization control based on frequency components cannot be performed and subjective image quality deteriorates
Solution Approach 1:
The patent segments the quantization process into two distinct stages: first applying zeroing out to specific frequency components (segmenting the frequency spectrum), then applying quantization matrices to the remaining coefficients. This segmentation allows selective control - high frequency components can be zeroed out for compression while low frequency components retain quantization control for quality preservation.
Solution Approach 2:
The patent applies different processing strategies to different frequency regions. Low frequency components (important for subjective quality) receive full quantization control with adjustable matrices, while high frequency components (less important) are subjected to zeroing out. This local quality differentiation resolves the contradiction by optimizing both compression and quality in their respective frequency domains.
2Ease of manufacture
If quantization matrices are applied to all orthogonal transform coefficients, then subjective image quality is improved, but code amount increases and compression efficiency deteriorates
Solution Approach 1:
The patent extracts and removes high frequency coefficients from the quantization process entirely by applying zeroing out before quantization matrices. This extraction eliminates the need to encode these coefficients, reducing code amount while focusing quantization resources on low frequency components that actually determine subjective image quality.
Solution Approach 2:
Instead of applying quantization matrices to all coefficients (excessive action), the patent applies them only to the necessary low frequency components (partial action). This partial application achieves the quality improvement goal without the penalty of increased code amount for encoding unnecessary high frequency data.
3Device complexity
If conventional quantization method with same-size matrix is used, then encoding simplicity is maintained, but adaptability to zeroing out technique and image quality improvement are limited
Solution Approach 1:
The patent introduces dynamic adaptability by making the quantization matrix size flexible rather than fixed. The quantization matrix size adapts to the block size and zeroing out configuration, allowing the system to handle different scenarios (different block sizes, different zeroing out patterns) while maintaining a relatively simple encoding framework based on standard quantization procedures.
Data Source
AI summary
A decoding unit decodes data corresponding to an N×M array of quantized coefficients from a bit stream. An inverse quantization unit derives orthogonal transform coefficients from the N×M array of quantized coefficients by using at least a quantization matrix. An inverse orthogonal transform unit performs inverse orthogonal transform on the orthogonal transform coefficients generated by the inverse quantization unit to generate prediction residuals corresponding to a block of a P×Q array of pixels. An inverse quantization unit derives the orthogonal transform coefficients by using at least a quantization matrix of an N×M array of elements, and the inverse orthogonal transform unit generates prediction residuals for the P×Q array of pixels having a size larger than the N×M array.