Automatic Thinning Detection through Image Segmentation Using Equivalent Array-Type Lamp-Based Lock-in Thermography.

Area of Science:

• Materials Science
• Non-Destructive Testing (NDT)
• Infrared Thermography

Background:

• Infrared thermography (IRT) is a reliable non-destructive testing technique for real-time thermal response measurement.
• Defect detection in materials is crucial for structural integrity and safety.

Purpose of the Study:

• To detect thinning defects in S275 specimens using lock-in thermography (LIT).
• To evaluate the automated detection and segmentation performance for defect characterization.

Main Methods:

• Lock-in thermography (LIT) was applied to S275 specimens with thinning defects.
• Four-point signal processing was used to acquire phase and amplitude images.
• Otsu algorithm for binarization and boundary tracking for automated defect detection.
• Root Mean Square Error (RMSE) was used to evaluate detectability.

Main Results:

• Optimal excitation frequency was determined for LIT.
• Automated detection and segmentation of thinning defects were achieved.
• Defect detectability was quantitatively evaluated using RMSE.

Conclusions:

• LIT is a viable technique for detecting thinning defects in S275 steel.
• Image segmentation and automated detection methods enhance defect clarification.
• RMSE provides a quantitative measure for evaluating defect detection performance.