Line Defect Detection Using Dual Illumination and Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting line defects on surfaces often result in false detections, particularly with edges being misclassified as true line defects, which hampers the accuracy of industrial quality assurance processes.
Innovation Solution
A system and method utilizing a combination of ambient and dark field illumination sources, image processing techniques, and machine learning classifiers to accurately detect line defects by capturing images, processing them to generate hypotheses, and refining line segments based on specific criteria to distinguish true defects from false ones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If edge detection is applied on an image of the surface to detect line defects, then defect detection capability is provided, but false detections increase (e.g., image borders detected as true line defects)
Solution Approach 1:
The patent segments the image processing into multiple stages: initial edge detection to identify potential defects, followed by separate verification steps including checking against image borders, analyzing line segment continuity, and applying machine learning classification. This multi-stage segmentation allows the system to filter out false detections (like image borders) while preserving true defect detection capability.
Solution Approach 2:
The patent introduces an intermediary machine learning classifier trained on characteristics of true defects as a mediator between raw edge detection results and final defect identification. This intermediary layer learns to distinguish true line defects from false detections (such as image borders) by analyzing patterns in the detected edges, thereby reducing false detection rates while maintaining detection accuracy.
2Measurement precision
If multiple image processing operations are performed to reduce false detections, then detection accuracy improves, but processing complexity increases
Solution Approach 1:
The complex image processing task is segmented into distinct modular operations: edge detection, border exclusion checks, line segment refinement, hypothesis generation, and machine learning classification. Each module performs a specific function and can be independently optimized or replaced, managing overall system complexity while achieving high detection accuracy through the coordinated sequence of these segmented operations.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The approach significantly reduces false detection rates and enhances the accuracy of line defect detection, enabling precise identification of even micro defects on surfaces, such as those found in the power generation industry.
Implementation Method 1
illuminating the surface of the object with a plurality of illumination sources comprising at least one ambient illumination source and at least one dark field illumination source
Data Source
AI summary
Method and system for detecting line defects on surface of object are presented. An imaging device captures images of surface of object under ambient and dark field illumination conditions. The images are processed with a plurality of image operations to detect areas of potential defects based on predictable pattern consisting of bright and shadow regions. Areas of potential defect are cut from processed images to sub images. Sub images are stitched together to generate hypotheses of potential defects at locations on surface of object. The hypotheses are classified to determine whether the potential defects are true defects at the locations. Line defect is detected by refining line segments detected on the processed image based on criteria. The criteria include distance from the true defects to the line segments and slops between the true defects and the line segments are less than threshold values.