Secondary Transform Coding for Low-Frequency Residual Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution and high-quality images leads to higher transmission and storage costs due to the increased amount of information, necessitating a more efficient image compression technique.
Innovation Solution
Implementing a non-separable secondary transform (NSST) index in image coding systems to enhance transform efficiency by concentrating non-zero transform coefficients on low-frequency components and optimizing residual coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution and high-quality images are transmitted or stored, then image quality is improved, but transmission cost and storage cost increase
Solution Approach 1:
The patent extracts only the essential visual information by applying transforms (DCT, DST, NSST) to convert image blocks into frequency domain coefficients, where most energy is concentrated in a few low-frequency coefficients. This allows discarding or coarsely encoding high-frequency coefficients, effectively extracting only the most important visual information while reducing data quantity.
Solution Approach 2:
The patent changes the representation parameters of image data by transforming spatial domain pixel values into frequency domain coefficients. This parameter transformation enables more efficient compression by exploiting the fact that human visual system is less sensitive to high-frequency variations, allowing aggressive compression of those components while preserving perceived image quality.
2Productivity
If conventional transform methods are used, then coding process is simple, but transform efficiency and residual coding efficiency are insufficient
Solution Approach 1:
The patent introduces dynamic adaptability by allowing the transform type to be selected and switched based on block characteristics, prediction mode, and other contextual factors. The encoder can dynamically choose among DCT, DST, and NSST transforms, and the decoder adapts accordingly, optimizing transform efficiency for different content types while managing complexity through conditional logic.
Solution Approach 2:
The patent segments the transform process into multiple stages and types. Instead of applying a single transform uniformly, it divides the coding process into different transform stages (primary transform, secondary NSST) and allows different transform types to be applied to different blocks based on their characteristics, improving overall transform efficiency while keeping each individual transform relatively simple.
3Productivity
If NSST is applied to concentrate non-zero coefficients on low-frequency components, then residual coding efficiency is improved, but decoding complexity increases
Solution Approach 1:
The patent applies NSST as a preliminary action before residual coding and entropy encoding. By pre-concentrating the energy into low-frequency coefficients through NSST, the subsequent residual coding process becomes more efficient because the transformed coefficients have better statistical properties and sparsity, making the overall process more efficient despite the additional transform step.
Solution Approach 2:
The NSST transform acts as an intermediary between the primary transform and the residual coding stage. It serves as a mediator that further processes the primary transform coefficients to enhance their compressibility, bridging the gap between spatial domain residuals and the final entropy-coded bitstream, thereby improving residual coding efficiency.
Data Source
AI summary
An image information decoding method performed by means of a decoding device according to the present invention comprises the steps of: decoding a non-separable secondary transform (NSST) index from a bitstream if NSST is applied to a target block; decoding information relating to transform coefficients with respect to the target block from the bitstream on the basis of the decoded NSST index; and deriving the transform coefficients with respect to the target block on the basis of the decoded information relating to the transform coefficients, wherein the NSST index is decoded prior to the information relating to the transform coefficients with respect to the target block.