Pixel-Adaptive Edge-Preserving Smoothing for Uneven Luminance Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional noise removal techniques using edge-preserving smoothing filters struggle to effectively remove noise from images with varying luminance distributions, leading to incomplete noise removal or blurring of edges.
Innovation Solution
An image processing method that calculates a relationship between luminance values and statistical values, sets parameters for an edge-preserving smoothing filter based on these values for each pixel, and applies the filter to the image, optimizing noise removal across the entire image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a fixed parameter edge-preserving smoothing filter is used, then the filter structure is simple, but noise removal effectiveness deteriorates in images with varying luminance distributions
Solution Approach 1:
The patent applies dynamics by transitioning from fixed parameters to dynamic parameters that adapt to local image characteristics. Specifically, the standard deviation σ is calculated locally for each pixel based on the luminance values in its neighborhood, allowing the filter parameters to change dynamically according to the local luminance distribution. This resolves the contradiction by making the filter adaptive to varying luminance regions while maintaining a relatively simple implementation framework.
Solution Approach 2:
The patent implements local quality by setting different filter parameters for different regions of the image based on local luminance statistics. Each pixel receives a customized standard deviation value calculated from its local neighborhood, ensuring that noise removal is optimized for each specific region's characteristics. This allows the filter to handle diverse luminance distributions across the image effectively.
2Manufacturing precision
If parameters are adjusted manually for different regions, then noise removal effectiveness improves, but operation complexity increases
Solution Approach 1:
The patent applies self-service by enabling the filter to automatically determine its own parameters based on the image content itself. The standard deviation for each pixel is calculated autonomously from the luminance values in its local neighborhood, without requiring external manual intervention. This resolves the contradiction by making the system self-adaptive, eliminating the need for manual parameter adjustment while achieving region-specific noise removal optimization.
Solution Approach 2:
The patent implements parameter changes by dynamically adjusting the standard deviation parameter based on local luminance statistics. Instead of using fixed or manually set parameters, the system automatically computes appropriate parameter values for each pixel based on the local image characteristics. This automated parameter adaptation improves noise removal effectiveness while maintaining ease of operation.
3Manufacturing precision
If strong filtering is applied to remove noise, then noise removal effectiveness improves, but edge preservation deteriorates
Solution Approach 1:
The patent applies local quality by using different filter strengths for different regions based on local luminance characteristics. In regions with high luminance variation (likely containing edges), the locally calculated standard deviation naturally results in gentler filtering. In uniform regions, stronger filtering can be applied. This spatially adaptive approach resolves the contradiction by applying appropriate filtering strength locally, removing noise effectively while preserving edges.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image acquisition device 1 includes a relationship calculation unit 202 that calculates relationship data between pixel values in an image and a statistical value of the pixel values, and a parameter setting unit 204 and a filter processing unit 205 that process an image to be processed. The parameter setting unit 204 and the filter processing unit 205 specify the statistical value for each pixel of the image on the basis of a pixel value for each pixel of the image and the relationship data, set parameters of an edge-preserving smoothing filter for each pixel of the image on the basis of the specified statistical value, and process the image using the edge-preserving smoothing filter in which the parameters are set for each pixel of the image.