Digital Image Processing for Ultrasonic Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing methods for digital images, particularly in ultrasonic non-destructive testing and other applications, are inefficient due to manual inspection of raw scan data, which can lead to incomplete defect detection and wasteful use of computing resources, as they rely on visual cues and are prone to noise and visualization issues.
Innovation Solution
An image processing method that samples images to generate patches, extracts feature vectors, clusters them, segments the image into regions, creates aggregate bounding regions, and filters out irrelevant data, automating the detection of regions of interest and enabling efficient analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual inspection of raw scan data is performed, then visual cues can be used to identify defects, but the process is time-consuming and may miss defects due to noise and visualization issues
Solution Approach 1:
The image is divided into multiple patches that are processed independently through clustering and segmentation operations, enabling parallel processing and automated defect identification while maintaining comprehensive coverage of the entire scan area
Solution Approach 2:
The manual visual inspection process is replaced with an automated computer-based system that uses image processing, clustering, and segmentation algorithms to identify defects, eliminating human subjectivity and enabling consistent, repeatable analysis
2Reliability
If the entire image is processed to detect defects, then comprehensive defect detection is achieved, but computing resources are wasted on irrelevant portions
Solution Approach 1:
The system extracts and processes only the relevant patches identified through clustering and segmentation operations, separating meaningful defect-containing regions from irrelevant background areas, thereby reducing computational workload while maintaining comprehensive defect detection
Solution Approach 2:
Instead of processing the entire image uniformly, the system applies processing operations only to selected patches that are likely to contain defects, using a partial processing approach that reduces overall computational resource consumption while maintaining detection effectiveness
Data Source
Figure 1~2b
Figure 3~5
Figure 4
AI summary
In an image processing method, for removing an unwanted portion of a digital image, the image, which contains a portion of interest, is sampled to generate a set of image patches representing the entire image. Feature vectors are extracted from the image patches, and the extracted feature vectors are partitioned into a plurality of clusters, where feature vectors in the same cluster share a common characteristic. The portion of interest in the image is segmented into a set of segments. An aggregate bounding region creation process is carried out, the process comprising: finding the largest segment in the set of segments and creating a bounding box around that segment; determining which image cluster contains the most image patches within the bounding box of the segment; and adding the image patches of the determined image cluster to an aggregate bounding region for the portion of interest. The aggregate bounding region creation process is repeated for each other segment in the set of segments in order of size, the resulting aggregate bounding region containing ail the image patches associated with the portion of interest. The image patches which fall outside the resulting aggregate bounding region are then removed from the image.