Single-Identifier Quantization Matrix Encoding for Image Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional quantization matrix encoding/decoding methods in image processing are limited in improving both objective and subjective image quality due to the lack of consideration for factors such as prediction mode, color component, size, form, one-dimensional transform type, and two-dimensional transform combination.
Innovation Solution
The method involves decoding and encoding a quantization matrix based on a single identifier derived from the current block's size, prediction mode, or color component, adjusting coefficient values, and encoding information on the quantization matrix, including subsampling and using default matrices when necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional quantization matrix encoding/decoding is used considering only prediction mode, color component, or size, then the encoding process is simple, but the objective image quality and subjective image quality are limited
Solution Approach 1:
The patent changes multiple parameters simultaneously including block size (4x4, 8x8, 16x16, 32x32), prediction mode (intra, inter, IBC), color component (luma, chroma), and transform type (DST, DCT) to create diverse quantization matrix configurations that improve image quality while maintaining encoding efficiency through systematic parameter variation
2Manufacturing precision
If multiple factors are considered for quantization matrix selection to improve image quality, then the encoding complexity increases
Solution Approach 1:
The patent performs preliminary classification by deriving a single identifier that encapsulates multiple factors (block size, prediction mode, color component, transform type) before quantization matrix selection. This preliminary identification step simplifies the encoding operation by pre-organizing the complex multi-factor decision space into a unified lookup structure
3Productivity
If a single identifier is derived to simplify quantization matrix selection, then the encoding efficiency improves, but the adaptability to different block characteristics may be reduced
Solution Approach 1:
The single identifier serves multiple functions simultaneously: it encodes block size information, prediction mode, color component type, and transform type. This multi-functional identifier maintains encoding efficiency while preserving adaptability to different block characteristics by consolidating multiple differentiation dimensions into one unified indexing mechanism
Data Source
AI summary
Disclosed herein is a method of encoding/decoding an image. The image decoding method includes decoding information on a quantization matrix from a bitstream, acquiring a quantization matrix of a current block based on the information on the quantization matrix, and dequantizing the current block using the quantization matrix of the current block. The acquiring of the quantization matrix of the current block includes deriving a single identifier using at least one of a size of the current block, a prediction mode or a color component, and acquiring the quantization matrix of the current block based on the single identifier.