Adaptive Color Denoising Filter Using Chrominance Weights
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional color denoising methods suffer from blurring of edges, high computational complexity, and luminance-biased performance due to reliance on edge detection and fixed thresholding, limiting their effectiveness in preserving color details and reducing noise.
Innovation Solution
A low-complexity color denoising method that computes adaptive weights based on chrominance level differences between pixels, using a spatial varying filter with a 17x17 pixel configuration and an approximation function to preserve edges without edge detection, allowing for efficient computation and flexible implementation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If low-pass filtering is applied to reduce color noise, then noise reduction is improved, but edge sharpness deteriorates
Solution Approach 1:
The patent applies different filtering strengths to different spatial locations by computing adaptive weights for each pixel based on local chrominance variance. Regions with high variance (edges) receive lower filtering strength, while regions with low variance (smooth areas) receive higher filtering strength, thus preserving edges while reducing noise in appropriate areas.
Solution Approach 2:
The filtering process transitions from static fixed-threshold filtering to dynamic adaptive filtering where the filtering weight for each pixel is dynamically computed based on local chrominance characteristics. This allows the filter to automatically adjust its behavior to match local image content, resolving the edge-blurring problem.
2Shape
If edge detection techniques are used to preserve edges during filtering, then edge sharpness is improved, but computational complexity increases
Solution Approach 1:
The patent extracts only the chrominance component for variance computation, ignoring the luminance component entirely. This selective extraction reduces computational load while still achieving edge preservation, as chrominance changes are sufficient to identify edges in color images.
Solution Approach 2:
The patent changes the parameter used for edge detection from traditional luminance-based gradients to chrominance-based variance. This parameter substitution simplifies the computational process while maintaining effective edge detection capability, reducing complexity without sacrificing edge sharpness.
3Manufacturing precision
If fixed thresholding is applied for noise suppression, then suppression capability is improved in certain directions, but adaptability to different image regions deteriorates
Solution Approach 1:
The patent transforms fixed thresholding into dynamic adaptive thresholding by computing a unique weight for each pixel based on its local chrominance environment. This allows the suppression capability to adapt automatically to different image regions, maintaining effectiveness across diverse content while preserving local details.
Data Source
AI summary
Methods and apparatuses for color denoising are described. There is provided an image processing method. The method comprises the steps of retrieving chrominance level for a plurality of pixels in a predetermined configuration; computing a weight for each pixel; applying the weight to the chrominance level for each the pixel; summing the weighted chrominance level to output a chrominance level for a center pixel at the center of said predetermined configuration; and repeating the above steps for every pixel per image. The color noise can be reduced while preserving the edge without any edge detection.