Phase-modulation combined deflectometry for small defect detection.

Area of Science:

• Optical metrology
• Computer vision
• Artificial intelligence

Background:

• Deflectometry is effective for specular surfaces but struggles with semispecular/diffuse surfaces due to low contrast and signal-to-noise ratio.
• Accurate defect detection on challenging surfaces remains a significant challenge in industrial inspection.

Purpose of the Study:

• To develop an accurate and robust defect detection method for semispecular and diffuse surfaces.
• To overcome the limitations of traditional deflectometry in low-contrast environments.

Main Methods:

• A phase-modulation combined method utilizing fringe phase and modulation was proposed.
• A dual-branch convolutional neural network (CNN) was employed to extract geometric and photometric features.
• The CNN processed phase-shifting pattern sequences and modulation data simultaneously.

Main Results:

• The proposed method demonstrated significantly improved defect detection performance.
• The dual-branch CNN effectively extracted relevant features for enhanced detection.
• The method showed a notable reduction in false positives compared to state-of-the-art techniques.

Conclusions:

• The phase-modulation combined method offers an effective solution for defect detection on challenging surfaces.
• The integration of CNNs with phase-modulation enhances accuracy and reliability in optical metrology.
• This approach holds promise for advancing automated inspection systems.