Signal Patch Denoising via Noise Orthogonal Neighbor Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional signal processing algorithms face a tradeoff between computational intensity and reduced signal quality, particularly in high-dimensional signals, where non-local algorithms like Non-Local Means (NLM) are computationally intensive and can introduce noise artifacts due to correlated noise between reference and neighbor patches.
Innovation Solution
A method that dynamically computes a reference distance based on the noise distribution of a signal patch to select neighbor patches, which are then processed to generate values for the reference patch, thereby maximizing noise orthogonality and improving noise cancellation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-local algorithms like NLM are used to process a high number of neighboring patches, then signal quality and noise cancellation are improved, but computational burden increases significantly
Solution Approach 1:
The algorithm segments the set of neighboring patches into two groups: those with orthogonal noise (useful for denoising) and those with correlated noise (discarded). This segmentation allows the algorithm to process only the relevant subset of patches, reducing computational burden while maintaining signal quality.
Solution Approach 2:
The algorithm applies different quality criteria to different neighboring patches based on their noise characteristics. Instead of uniformly processing all neighboring patches, it selectively processes only those patches whose noise is orthogonal to the reference patch, optimizing the local quality of noise cancellation.
2Productivity
If nearest neighbor approach is used to limit the number of neighboring patches, then computational burden is reduced, but noise cancellation quality deteriorates due to correlated noise
Solution Approach 1:
The algorithm changes the selection parameter from simple spatial proximity (nearest neighbor) to noise orthogonality. By computing the noise distribution and selecting patches based on orthogonal noise characteristics rather than just distance, the algorithm maintains noise cancellation quality while reducing the number of patches processed.
3Device complexity
If neighboring patches with correlated noise are used for processing, then computational complexity is reduced, but artifacts are introduced in the processed signal
Solution Approach 1:
The algorithm converts the potentially harmful correlated noise into a useful selection criterion by identifying and excluding patches with correlated noise. By using noise orthogonality as the selection basis, the algorithm transforms the noise characteristic from a harmful factor into a beneficial filter that prevents artifact introduction.
Data Source
AI summary
A system, method and computer program product are provided for generating one or more values for a signal patch using neighboring patches collected based on a distance dynamically computed from a noise distribution of the signal patch. In use, a reference patch is identified from a signal, and a reference distance is computed based on a noise distribution in the reference patch. Neighbor patches are then collected from the signal based on the computed reference distance from the reference patch. Further, the collected neighbor patches are processed with the reference patch to generate one or more values for the reference patch.