Adaptive Denoising Using Internal and External Patches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image denoising techniques face challenges in effectively reducing noise in digital images, particularly when there is a limited number of self-similar patches, leading to artifacts or blurring, and struggle with images having only small or limited noise.
Innovation Solution
An adaptive denoising method that utilizes both internal and external patches by grouping external image patches into partitions, determining partition center patches, and generating denoising operators based on clean-noisy patch pairs, classifying image patches as common or complex, and applying appropriate denoising operators to reduce noise effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If internal patches are used for denoising, then denoising can be performed without external references, but artifacts or blurring occur when self-similar patches are insufficient
Solution Approach 1:
The patent combines internal patches from the noisy image with external patches from example images to form a unified denoising approach. When internal patches are insufficient, external patches supplement the denoising process, preventing artifacts and blurring while maintaining image quality.
Solution Approach 2:
The patent dynamically adapts between using internal patches and external patches based on the availability and quality of self-similar patches in the image. The system transitions between denoising modes depending on the specific image content and noise characteristics.
2Adaptability or versatility
If external patches are used for denoising, then denoising can be performed with limited internal similarity, but performance degrades for images with small or limited noise
Solution Approach 1:
The patent implements dynamic adaptation by evaluating the noise level and patch similarity in the input image, then adjusting the reliance on external versus internal patches accordingly. For images with small or limited noise, the system reduces dependence on external patches to maintain reliability.
Solution Approach 2:
The system uses feedback from patch comparison metrics and noise level assessment to determine the optimal denoising strategy. The quality of match between internal and external patches provides feedback that guides whether to prioritize internal or external patch-based denoising.
3Manufacturing precision
If all partitions are retained for denoising, then comprehensive coverage is achieved, but computational complexity increases
Solution Approach 1:
The patent extracts and removes partitions with fewer similar example patches, retaining only the most relevant partitions for denoising. This extraction process reduces computational complexity by eliminating redundant partitions while preserving denoising accuracy for the most important image regions.
Solution Approach 2:
The patent segments the set of partitions into retained and removed groups based on the number of similar example patches. This segmentation allows selective processing of only the most valuable partitions, reducing overall computational complexity while maintaining denoising quality.
Data Source
AI summary
In techniques for adaptive denoising with internal and external patches, example image patches taken from example images are grouped into partitions of similar patches, and a partition center patch is determined for each of the partitions. An image denoising technique is applied to image patches of a noisy image to generate modified image patches, and a closest partition center patch to each of the modified image patches is determined. The image patches of the noisy image are then classified as either a common patch or a complex patch of the noisy image, where an image patch is classified based on a distance between the corresponding modified image patch and the closest partition center patch. A denoising operator can be applied to an image patch based on the classification, such as applying respective denoising operators to denoise the image patches that are classified as the common patches of the noisy image.