Image Encoding Reducing Ring Noise via Adaptive Quantization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High compression rates in image encoding lead to severe image distortion due to ring noise and fixed-pattern noise, affecting subjective image quality.
Innovation Solution
A data block encoding method that determines the presence of intensive edge regions and high frequency noise, adjusting quantization parameters and dividing data blocks into smaller coding units to minimize distortion, thereby reducing ring and fixed-pattern noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a high compression rate is used in image encoding, then the compression efficiency is improved, but image distortion due to ring noise and fixed-pattern noise increases
Solution Approach 1:
The patent applies local quality by differentiating encoding strategies for different regions within an image block. It identifies intensive edge regions separately from regular regions and applies different quantization parameters to each. This allows the encoder to maintain higher quality in distortion-prone edge regions while using more aggressive compression in regular regions, thus resolving the contradiction between compression efficiency and image distortion.
Solution Approach 2:
The patent segments the image block into multiple coding units (CUs) of different sizes based on the presence of intensive edge regions. By dividing the block into smaller CUs in regions with edge content, it can apply more fine-grained control over quantization and transform processes, reducing ring noise and fixed-pattern noise while maintaining overall compression efficiency.
2Productivity
If a large DCT transform size is used, then the compression rate increases, but ring phenomenon becomes severer and subjective image quality is affected
Solution Approach 1:
The patent implements dynamic transform block size selection based on the content characteristics of each region. Instead of using a fixed large transform size for all blocks, it adapts the transform block size to match the coding unit size, which varies according to the presence of intensive edge regions. This dynamic adjustment reduces ring phenomenon in edge regions while maintaining compression efficiency in regular regions.
Solution Approach 2:
The patent applies different transform block sizes to different regions of the image based on local characteristics. In regions containing intensive edge regions, smaller transform blocks are used to suppress ring noise, while larger transform blocks are used in regular regions to maintain compression efficiency. This local adaptation resolves the contradiction between compression rate and ring phenomenon.
3Productivity
If quantization is performed to cancel high frequency coefficients, then image compression is achieved, but noise intensity increases gradually with compression rate
Solution Approach 1:
The patent applies different quantization parameters (QPs) to different regions of the image block. In intensive edge regions, a reduced QP is used to preserve high frequency information and reduce fixed-pattern noise, while in regular regions, a higher QP is used to achieve compression. This local differentiation allows image compression to be achieved while controlling noise intensity in distortion-prone regions.
Data Source
AI summary
A data block encoding method and apparatus are provided. The data block encoding method includes: determining whether a data block includes an intensive edge region, where the intensive edge region is a region including an image distortion generated by ring noise, and the data block is a data block in a to-be-encoded image; and when the data block includes an intensive edge region, reducing a value of a quantization parameter used for encoding the data block, and encoding the data block by using a reduced value of the quantization parameter, or dividing the data block into multiple coding units of different sizes according to different coding layers, adjusting rate-distortion costs of the multiple coding units of different sizes, and encoding the data block by using a coding unit with a minimum rate-distortion cost obtained after the adjustment.