Steel Surface Defect Detection Using Multi-Directional Bright Overlap
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately distinguish between concave-convex surface defects and base steel portions, which have similar reflectance characteristics, leading to incorrect detection and reduced manufacturing yield in steel materials.
Innovation Solution
A detecting method and apparatus that utilize distinguishable illumination from symmetric light sources and image processing to identify overlapping bright portions in two-dimensional images, enabling accurate differentiation between base steel portions and surface defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simple image capturing is used to detect surface defects, then the detection process is simple and fast, but the ability to distinguish between harmful concave-convex defects and harmless flat scale patterns is insufficient
Solution Approach 1:
The detection process is segmented into multiple stages: initial simple image capturing for fast screening, followed by subtraction image processing to enhance specific defect features, and finally machine learning-based classification. This multi-stage segmentation allows the system to maintain high detection speed while improving classification accuracy through progressive filtering and analysis.
Solution Approach 2:
The invention transitions from two-dimensional image intensity analysis to three-dimensional surface shape analysis by calculating height information from stereo images. This dimensional transformation enables the system to distinguish between flat scale patterns (zero height variation) and concave-convex defects (non-zero height variation), significantly improving classification accuracy without sacrificing detection speed.
2Measurement precision
If subtractive image processing is applied to distinguish concave-convex defects from flat scale patterns, then classification accuracy is improved, but base steel portions with high reflectance are misidentified as concave-convex defects
Solution Approach 1:
The invention introduces height information as a new dimension to differentiate between base steel portions and concave-convex defects. While both may appear similar in two-dimensional subtraction images, base steel portions have zero height variation whereas concave-convex defects exhibit non-zero height variations. This dimensional addition eliminates false positives without reducing detection accuracy.
Solution Approach 2:
The invention introduces machine learning models as an intermediary between subtraction image generation and final defect classification. The trained model learns to recognize the subtle differences between base steel portions and actual defects by analyzing multiple features including height information, texture patterns, and spatial relationships, thereby reducing misidentification while maintaining high accuracy.
3Measurement precision
If multiple light beams from different directions are used to capture images, then the ability to generate subtraction images for defect detection is improved, but the device complexity and processing time increase
Solution Approach 1:
The illumination system is segmented into two functional groups: multiple light sources for generating subtraction images to enhance defect visibility, and a separate stereo imaging system for capturing height information. This segmentation allows each subsystem to be optimized independently, managing overall system complexity while maintaining high detection precision through coordinated operation of the components.
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
Enhances the ability to distinguish between base steel portions and surface defects, improving manufacturing yield by reducing incorrect detections and enhancing quality management in steel material production.
Implementation Method 1
area sensors capture two-dimensional images formed by reflected light beams of the illumination light beams
Data Source
Figure 1
Figure 2(a)~2(e)
Figure 3(a)~4(e)
AI summary
A surface-defect detecting method according to the present invention is a surface-defect detecting method of optically detecting a surface defect of a steel material and includes an irradiation step of irradiating a same examination target part with illumination light beams from different directions by using two or more distinguishable light sources; and a detection step of detecting a surface defect in the examination target part based on the degree of overlapping of bright portions extracted from two or more images formed by reflected light beams of the illumination light beams.