Image Denoising via Edge-Constrained Template Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The computational complexity of non-local means (NLM) denoising filters is high, leading to increased computation time in decoding processes, particularly in video coding standards, which hinders their practical application due to the need for template matching across all pixels within arbitrary search shapes.
Innovation Solution
The proposed solution reduces computational complexity by limiting the search shape and template shape based on edge directions and deviation degrees, focusing search points along edge directions and reducing the number of comparisons, thereby reducing the number of calculations required for template matching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If template matching is performed across all pixels within arbitrary search shapes, then denoising quality is improved, but computational complexity increases
Solution Approach 1:
The patent segments the search space by dividing it into edge regions and non-edge regions based on edge detection results. Different search shapes are applied to different regions: a first search shape for edge pixels and a second search shape for non-edge pixels. This segmentation allows the algorithm to perform comprehensive template matching in non-edge regions for high denoising quality while limiting the search scope in edge regions to reduce computational complexity.
Solution Approach 2:
The patent applies different search shapes and template matching strategies to different local regions of the image. Edge pixels use a constrained search shape that limits the number of search points, while non-edge pixels use a more comprehensive search shape. This local differentiation ensures that denoising quality is maintained in non-edge regions while computational complexity is reduced in edge regions, resolving the contradiction between quality and complexity.
2Manufacturing precision
If template matching is performed across all pixels within arbitrary search shapes, then denoising quality is improved, but processing time increases
Solution Approach 1:
The patent segments the image into edge and non-edge regions, applying different processing strategies. In edge regions, the search shape is constrained to reduce the number of template matching operations, thereby reducing processing time. In non-edge regions, full template matching is performed to maintain denoising quality. This segmentation resolves the contradiction by optimizing processing time in critical regions without sacrificing overall quality.
Solution Approach 2:
The patent applies partial template matching in edge regions where full matching would be excessive and computationally expensive. By using a constrained search shape that includes fewer search points, the algorithm performs a reduced but sufficient level of matching for edge pixels, while performing complete matching for non-edge pixels. This partial action approach maintains acceptable denoising quality while significantly reducing processing time.
Data Source
AI summary
In an image processing apparatus, an edge direction detection unit calculates an edge direction using a denoising target image. A search shape setting unit sets a search shape in which the number of search points along the direction perpendicular to the edge direction is smaller than the number of search points along the edge direction, using the calculated edge direction as an index used for limiting the search shape. When noise of a target pixel is removed with a weight in accordance with the degree of template similarity between a template for the target pixel and a template for each of search points within a search shape and the weighted sum of pixel values at the search points, a filter execution unit executes a process of performing template matching on only search points within the set search shape and removing the noise of the target pixel.