Steel Surface Defect Detection With Dual-Angle Illumination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to distinguish between scales or harmless patterns and surface defects on steel materials with high precision due to varying reflectance and numerous defect types, making accurate detection difficult.
Innovation Solution
A surface defect detection apparatus and method using two irradiation units with different illumination angles and imaging units to capture and process two-dimensional images, performing difference processing to remove harmless patterns and scales, and detecting surface defects based on shape features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If light reflection method is used to detect surface defects, then detection speed is improved, but false detection of scales and harmless patterns increases
Solution Approach 1:
The invention segments the detection process into multiple stages: first capturing images with oblique illumination to highlight surface topology, then using image processing to identify and remove scales and harmless patterns based on their visual characteristics, and finally detecting actual defects in the processed images. This segmentation allows rapid processing while maintaining accuracy by treating different surface features at appropriate stages.
Solution Approach 2:
The invention introduces an intermediary image processing step that acts as a mediator between raw image capture and defect detection. This processing stage analyzes image characteristics to distinguish scales and harmless patterns from actual defects, preventing false detections while preserving true defect signals for final identification.
2Adaptability or versatility
If multiple image combination method is used to detect surface defects, then detection coverage is improved, but system complexity and detection logic difficulty increase
Solution Approach 1:
The invention uses segmentation to divide the detection task into distinct phases: oblique illumination imaging for topology enhancement, image processing for pattern recognition and filtering, and defect detection in processed images. This segmentation manages complexity by handling different aspects of detection separately rather than simultaneously.
Solution Approach 2:
The invention applies preliminary action by performing image processing operations before final defect detection. The system pre-processes images to remove scales and harmless patterns, preparing the data in advance for more accurate and simpler defect identification in the subsequent stage.
3Device complexity
If simple light reflection detection is used, then device simplicity is maintained, but ability to distinguish defect types deteriorates
Solution Approach 1:
The invention applies partial action by implementing image processing operations selectively focused on removing scales and harmless patterns rather than attempting to analyze all image features. This targeted approach maintains relative device simplicity while significantly improving defect discrimination capability through selective processing.
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
Enables precise differentiation between scales or harmless patterns and surface defects, enhancing the accuracy of defect detection on steel materials.
Implementation Method 1
applies illumination light 12a to an inspection target part 11... receives reflected light of the illumination light 12a
Data Source
Figure 1
Figure 2
Figure 3
AI summary
First illumination light and second illumination light that are distinguishable from each other are each applied to an inspection target part of a surface of a steel material with substantially the same incident angle from directions inclined opposite to each other. A first image of the inspection target part illuminated with first illumination light and a second image of the inspection target part illuminated with second illumination light are each captured. A difference image between the first image and second image is generated. A combination of a bright part and a dark part of a protruding part in the inspection target part is removed from among bright parts and dark parts of the difference image based on an arrangement of a bright part and a dark part in a predetermined direction corresponding to the irradiation direction of the first or second illumination light. The presence or absence of a recessed part in the inspection target part is determined based on a shape feature amount of the remainder of the bright parts and the dark parts after this removal processing or an arrangement of the remainder of the bright parts and the dark parts in the predetermined direction. The recessed part that has been determined to be present is detected as a surface defect of the steel material.