Image Processing Edge Segmentation for Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Medical image processing often results in overshoot or undershoot artifacts due to edge enhancement, particularly when strong edges are present, which degrade image quality and can render images unusable for diagnosis.
Innovation Solution
An image processing method that identifies weak and strong edges in an input image, filters them using filters of different smoothness levels, and generates an output image based on edge information and a smoothed image to reduce artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If edge enhancement is performed on the original image to display more details, then image detail visibility is improved, but overshoot or undershoot artifacts appear on the image
Solution Approach 1:
The patent segments edges into different types (strong edges and weak edges) based on their gradient characteristics. By identifying and categorizing edges differently, the processing can be tailored to each type, preventing artifacts on strong edges while preserving details on weak edges.
Solution Approach 2:
The patent applies different filtering strengths to different regions of the image based on local edge characteristics. Strong edges receive stronger smoothing to prevent artifacts, while weak edges receive milder smoothing to preserve details. This local differentiation resolves the contradiction between detail visibility and artifact prevention.
2Object-generated harmful factors
If a stronger filter is used to reduce artifacts on strong edges, then artifact reduction is improved, but details on weak edges are lost
Solution Approach 1:
The patent segments edges into strong edges and weak edges based on gradient magnitude thresholds. This segmentation allows the system to apply appropriate filtering strength to each segment, using stronger filters for strong edges to reduce artifacts and weaker filters for weak edges to preserve details.
Solution Approach 2:
The patent implements local quality by adjusting filter parameters based on local edge strength characteristics. Regions with strong edges receive aggressive smoothing to eliminate artifacts, while regions with weak edges receive gentle smoothing to maintain detail visibility, thus resolving the contradiction.
3Device complexity
If uniform filtering is applied to the entire image, then processing simplicity is maintained, but both strong edges and weak edges are filtered with the same smoothness causing loss of detail
Solution Approach 1:
The patent introduces segmentation based on edge strength classification, dividing the image processing into different pathways for strong edges and weak edges. This segmentation enables differentiated filtering strategies while maintaining relatively simple processing logic through threshold-based classification.
Solution Approach 2:
The patent applies local quality by using different filter smoothness parameters for different regions. The system automatically adjusts filtering strength based on local edge characteristics, ensuring that strong edges are smoothed adequately while weak edges preserve their detail, thus improving detail preservation without excessive complexity.
Data Source
AI summary
An image processing method comprises: identifying a weak edge comprising a plurality of weak edge pixels and a strong edge comprising a plurality of strong edge pixels in an input image; filtering at least a part of said input image to obtain a smoothed image, during which said weak edge in said input image is filtered with a first filter and said strong edge in said input image is filtered with a second filter having a smoothness less than that of said first filter; acquiring edge information of said input image based on said input image and said smoothed image; and generating an output image based on said edge information and said smoothed image. The present invention further relates to an image processing system and an imaging system.