Transform-Based Image Coding with Adaptive LFNST Parsing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution and high-quality images/videos, including immersive media, leads to higher transmission and storage costs due to increased bit amounts, necessitating a more efficient image/video compression technique.
Innovation Solution
An image decoding method that includes receiving residual information with transform skip flags for each color component, setting a variable based on the presence of significant coefficients, parsing an LFNST index, and deriving an LFNST kernel to enhance coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If high-resolution and high-quality images/videos are transmitted or stored, then image quality and resolution are improved, but transmission cost and storage cost increase due to increased bit amount
Solution Approach 1:
The patent changes the parameter of transform coefficient representation by introducing LFNST (Low Frequency Non-Separable Secondary Transform) to transform residual signals. This transforms the frequency domain coefficients into a different basis that achieves better energy compaction, thereby improving compression efficiency while maintaining image quality. The transform skip flag allows dynamic selection between different transform parameters based on block characteristics.
Solution Approach 2:
The patent introduces dynamic control mechanisms including transform skip flags for each color component and LFNST index coding that adapts to different block types (single tree vs. dual tree luma). This dynamic adaptation allows the system to optimize the transform parameters based on local image characteristics, achieving better compression efficiency without sacrificing image quality in different regions.
2Productivity
If conventional transform methods are used for compression, then implementation is simple, but coding efficiency is insufficient for high-resolution images
Solution Approach 1:
The patent segments the transform processing by separating the primary transform from the secondary LFNST processing. The transform skip flag enables selective application of LFNST on a per-block basis, and the dual-tree structure allows independent processing of luma and chroma components. This segmentation reduces overall complexity by avoiding unnecessary transforms on blocks where they would not improve coding efficiency.
Solution Approach 2:
The patent applies LFNST selectively rather than universally through the transform skip flag mechanism. When the flag indicates transform skipping is appropriate, the computationally intensive LFNST is omitted. This partial application of the transform achieves coding efficiency improvements where needed while avoiding the complexity overhead in regions where conventional transforms suffice.
Data Source
AI summary
An image decoding method according to the present document comprises the steps of: receiving residual information in a bitstream, wherein the residual information includes a transform skip flag for each individual component of the current block; deriving corrected transform coefficients, wherein the step for deriving includes a step for setting a variable indicating whether effective coefficients exist at positions other than that of a DC component of the current block to 0 when even one of the values of the transform skip flags for the individual components is 0, and a step for parsing an LFNST index on the basis of the variable being 0; and deriving an LFNST kernel for applying LFNST on the basis of the LFNST index, wherein the variable being 0 may indicate that effective coefficients exist at positions other than that of the DC component.