Image Processing Unit Adaptive Filtering for Edge Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods, such as simple filtering and ε-filters, often deteriorate image quality by blurring edge portions when attempting to reduce noise, especially during repeated filtering processes.
Innovation Solution
An image processing unit that determines a candidate pixel group based on divergence in position and pixel value, associates these pixels with a pixel of interest, and performs image processing to smooth the pixel values while maintaining edge sharpness through controlled repeat counts, ensuring that edge portions are retained even after multiple filtering iterations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If repeated filtering is performed to reduce noise, then noise reduction effect is improved, but edge portions become blurred and expanded
Solution Approach 1:
The patent applies dynamics by making the filter processing adaptive rather than static. The processing unit dynamically adjusts which pixels are filtered based on local image characteristics (edge detection), so that filtering strength varies across different regions of the image. This allows repeated filtering to be performed on non-edge regions while preserving edge regions, resolving the contradiction between noise reduction and edge sharpness.
Solution Approach 2:
The patent implements local quality by applying different processing treatments to different regions of the image. Edge portions are identified and given special treatment (preserved from filtering), while non-edge regions undergo full filtering. This localized approach ensures that noise reduction is applied where safe, while edge regions maintain their sharpness, thus resolving the contradiction between these two competing requirements.
2Object-affected harmful factors
If simple filtering is applied to remove noise, then noise components are reduced, but important image portions such as edges are lost
Solution Approach 1:
The patent applies preliminary action by performing edge detection and identification before the filtering process. The processing unit first identifies which pixels constitute edge portions, then uses this information to guide the subsequent filtering operation. This preliminary classification ensures that edge information is protected from filtering before any noise reduction occurs, preventing loss of important image features.
Solution Approach 2:
The patent implements local quality by applying different filtering treatments to different regions. Edge regions are identified and excluded from filtering, while non-edge regions undergo noise reduction. This selective approach removes noise components from safe regions while preserving edge information, thus resolving the contradiction between noise reduction and information retention.
3Productivity
If ε-filter is repeatedly used to filter images, then filtering can be applied to entire image or divided blocks, but edge pixels and neighboring pixels become similar causing expanded and blurred edges
Solution Approach 1:
The patent applies dynamics by making the filtering process adaptive based on local image characteristics. Rather than uniformly applying ε-filter to all pixels, the processing unit dynamically determines which pixels are edges and adjusts filtering accordingly. This dynamic adaptation allows comprehensive filtering coverage on non-edge regions while protecting edge regions, resolving the contradiction between productivity and edge definition.
Solution Approach 2:
The patent implements local quality by applying different filtering strategies to different regions. Edge pixels are identified and given special treatment (preserved from filtering), while neighboring non-edge pixels undergo filtering. This localized approach maintains edge definition while still providing noise reduction in appropriate regions, thus resolving the contradiction between filtering coverage and edge definition.
Data Source
AI summary
An image processing unit includes a candidate pixel group determiner to specify each pixel of an image to be processed as pixel of interest, extract pixels having divergence in position in a certain range from the pixel of interest, and determine, from the extracted pixels, pixels having divergence in pixel value in a certain range from the pixel of interest as a candidate pixel group, an associator to associate the candidate pixel group with the pixel of interest, and an image processor to perform image processing on the candidate pixel group associated with the pixel of interest. The image processor selects, from the candidate pixel group, pixels having divergence in pixel value in a certain range from the pixel of interest as a pixel group to be processed, and smoothes the pixel value of the pixel of interest in accordance with each pixel value of the candidate pixel group.