Image Noise Filter Using Edge-Based Adaptive Wavelet Thresholding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing methods, such as spatially adaptive color correction and wavelet de-noising, fail to effectively reduce chrominance noise and artifacts like the Gibbs phenomenon and checkerboard reconstruction in color imaging, especially in low light conditions.
Innovation Solution
An image noise filter comprising a wavelet transform module and an edge-based adaptive filter module that uses dual tree wavelet analysis and synthesis, along with adaptive thresholding based on edge information and local variance, to filter high-frequency chrominance and luminance sub-bands in a modified hue-saturation chromaticity space, reducing chrominance noise and artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If wavelet de-noising is applied to reduce chrominance noise, then noise reduction is improved, but ringing artifacts and checkerboard reconstruction artifacts are introduced
Solution Approach 1:
The patent applies different thresholding strategies to different regions of the wavelet coefficients based on edge detection. High-frequency coefficients near edges are preserved while coefficients in flat regions are more aggressively thresholded. This local differentiation allows noise reduction in flat areas without introducing artifacts in edge regions, resolving the contradiction between noise reduction and artifact generation.
Solution Approach 2:
The patent uses adaptive thresholding where the threshold value is dynamically adjusted based on local image characteristics (edge presence, local variance). The threshold is not fixed but adapts to the specific region being processed, allowing optimal noise reduction while preserving edges and avoiding artifact generation in different image regions.
2Object-affected harmful factors
If spatially adaptive color correction matrix is used to reduce chrominance noise, then chrominance noise is reduced, but post processing remains problematic
Solution Approach 1:
The patent combines multiple processing steps (wavelet transform, edge detection, adaptive thresholding, inverse wavelet transform) into a unified de-noising framework. By merging these operations into a single integrated algorithm rather than separate sequential processing steps, the complexity of post-processing is reduced while maintaining effective chrominance noise reduction.
Solution Approach 2:
The algorithm uses the image data itself (through wavelet coefficients and edge detection) to automatically determine processing parameters and thresholds. The system self-adjusts based on the input image characteristics without requiring complex external control or multiple processing passes, simplifying the overall post-processing workflow.
3Object-affected harmful factors
If bilateral filtering or anisotropic diffusion is applied to de-noise images, then noise reduction is achieved, but computational complexity and processing time increase
Solution Approach 1:
The patent segments the image processing into distinct stages: wavelet transform, edge detection, adaptive thresholding, and inverse wavelet transform. This segmentation allows each stage to be optimized independently and enables parallel processing of different image regions, significantly reducing overall processing time while maintaining effective noise reduction comparable to more computationally intensive methods like bilateral filtering.
Data Source
AI summary
An image noise filter includes a wavelet transform module and an edge based adaptive filter module. The dual tree wavelet transform module provides low frequency wavelet information and high frequency wavelet information in response to image information. The edge based adaptive filter module provides filtered high frequency wavelet information in response to the high frequency wavelet information and edge information that is based on the low frequency wavelet information.