Multi-scale Image Noise Reduction via Local Variance Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods struggle to effectively reduce noise in digital images, particularly in medical radiographic images, while preserving edge contrast and avoiding distortions near significant grey level transitions.
Innovation Solution
A method that decomposes a digital image into detail images at multiple resolution levels and a residual image, processes the detail images by calculating local variance for each pixel, and applies a reconstruction algorithm to enhance or suppress pixels based on their local variance and value, thereby improving the signal-to-noise ratio without reducing edge contrast.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If linear filters are used to reduce noise, then noise attenuation is improved, but edge sharpness and small structure preservation deteriorate
Solution Approach 1:
The patent divides the image processing task into multiple resolution levels using a pyramid structure. At each level, noise reduction is applied independently, allowing different processing strengths at different scales. This segmentation enables noise attenuation in smooth regions while preserving edges through the multi-scale decomposition.
Solution Approach 2:
The patent applies local variance calculation to determine processing parameters at each pixel location. By adapting the noise reduction strength based on local image statistics (variance), the filter automatically preserves edges (high variance) while reducing noise in homogeneous regions (low variance), achieving local optimization of both noise attenuation and edge preservation.
2Manufacturing precision
If non-linear filters are used to preserve edges, then edge sharpness is improved, but noise attenuation capability deteriorates
Solution Approach 1:
The multi-resolution pyramid decomposes the image into multiple levels, where coarse levels capture broad structures and fine levels capture details and edges. By processing at multiple levels, the patent achieves both noise reduction at coarser levels and edge preservation at finer levels, combining the advantages of both linear and non-linear approaches.
Solution Approach 2:
The patent uses local variance to adaptively control the filtering strength. In regions with high local variance (edges), the filter applies minimal smoothing to preserve sharp transitions. In regions with low local variance (noise-prone homogeneous areas), the filter applies stronger noise reduction, thus achieving both edge preservation and noise attenuation.
3Object-affected harmful factors
If larger filter sizes are used to handle broader noise bands, then noise reduction effectiveness is improved, but computational cost and processing time increase
Solution Approach 1:
The patent segments the frequency spectrum into multiple resolution levels in the pyramid structure. Each level handles a specific frequency band, allowing the use of smaller, computationally efficient filters at each level rather than requiring one large filter to handle the entire noise spectrum. This divides the computational burden across multiple smaller processing stages.
Solution Approach 2:
The patent transforms the single-stage spatial filtering problem into a multi-stage hierarchical processing problem across resolution levels. By adding the resolution level dimension, the patent achieves broad noise band coverage through cumulative processing at multiple levels while keeping individual filter operations small and efficient.
4Object-affected harmful factors
If adaptive noise filtering with local statistics is applied, then noise reduction effectiveness is improved, but computational complexity increases
Solution Approach 1:
The patent segments the image into a pyramid structure with multiple resolution levels. By computing local variance at each level independently, the computational complexity is distributed across levels rather than requiring a single complex pass over the entire image. This hierarchical segmentation reduces the overall computational burden while maintaining adaptive noise filtering effectiveness.
Solution Approach 2:
The patent adds the resolution level dimension to the processing hierarchy. Local statistics (variance) are computed at each level, allowing adaptive filtering to operate on downsampled representations at coarser levels, which reduces the number of pixels requiring full computational treatment and thus lowers overall complexity.
Data Source
AI summary
This invention is related to a multi-scale image processing algorithm for enhancing contrast to noise ratio in an electronic representation of an image represented by a) decomposing said digital image into a set of detail images at multiple resolution levels and a residual image at a resolution level lower than said multiple resolution levels, b) processing said detail images, c) computing a processed image by applying a reconstruction algorithm to the residual image and the processed detail images, said reconstruction algorithm being such that if it were applied to the residual image and the detail images without processing, then said digital image or a close approximation thereof would be obtained. The processing comprises the steps of: d) calculating a local variance value for at least one detail image pixel within a neighbourhood of said detail image pixel; e) enhancing or suppressing at least one detail image pixel as a function of said local variance value and said detail image pixel value.