Image Noise Reduction Using Adaptive Weight Derivation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The non-local means method for image noise reduction is insufficient due to the rare occurrence of complete block matches, leading to high weights for the target pixel and insufficient noise reduction.
Innovation Solution
An image processing apparatus that determines pixel values by a weighted mean of plural reference pixels, using a derivation unit to derive weights based on similarity between target and reference regions, and a weighted mean processing unit that performs weighted mean processing using a pixel other than the target pixel as the reference pixel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the non-local means method is used for noise reduction, then noise reduction processing is performed, but complete block matches occur rarely leading to insufficient noise reduction effect
Solution Approach 1:
The patent applies local quality by differentiating the treatment of reference pixels based on their spatial relationship with the target pixel. Specifically, it excludes the target pixel itself from being a reference pixel, and applies different weight calculation methods for reference pixels at different locations. This localized differentiation ensures that self-similarity (complete block match) does not dominate the noise reduction process, thereby improving noise reduction effectiveness.
Solution Approach 2:
The patent inverts the conventional approach by not using the target pixel as a reference pixel, which is counterintuitive since self-comparison would provide perfect similarity. Instead, it deliberately excludes the target pixel from reference pixel candidates, forcing the system to rely on surrounding pixels for noise reduction, thus avoiding the trivial solution of complete block match.
2Reliability
If simple noise reduction processing is applied, then noise is reduced, but resolution deteriorates
Solution Approach 1:
The patent applies local quality by calculating weights based on the specific spatial relationship between reference pixels and the target pixel. Reference pixels at different locations contribute differently to the weighted mean, with pixels having higher similarity and appropriate spatial proximity receiving higher weights. This localized differentiation preserves edges and textures while reducing noise.
Solution Approach 2:
The patent changes the parameter of weight assignment by using a weighted mean approach where weights are determined by similarity measures and spatial relationships. Instead of uniform averaging, the system dynamically adjusts weights based on pixel characteristics, achieving noise reduction without blurring edges or losing resolution.
Data Source
AI summary
A non-local means method is insufficient in its noise reduction effect or edge retainability due to a perfect match between blocks in a case where a reference pixel matches a target pixel. Therefore, information on a target region and plural reference regions is obtained for the target pixel. Whether the target region matches any one of the reference regions is determined from the obtained information. Switching between weight derivation methods based on similarity between the target region and the reference region is performed according to a determined result.