Automated defect detection and classification for fiber-optic coil based on wavelet transform and self-adaptive...

Area of Science:

• Materials Science
• Optical Engineering
• Computer Vision

Background:

• Manual quality monitoring of fiber-optic coils (FOCs) in winding systems is inefficient and prone to inaccuracies.
• Automated defect detection is crucial for improving the reliability and consistency of FOC production.

Purpose of the Study:

• To develop a reliable machine vision-based framework for automatic defect detection in fiber-optic coils (FOCs).
• To address the limitations of manual inspection methods in terms of efficiency and accuracy.

Main Methods:

• A defect detection scheme integrating wavelet transform and nonlocal means filtering was employed for precise defect localization.
• A support vector machine (SVM) classifier was utilized, with parameters optimized by a self-adaptive genetic algorithm.
• An in-house imaging system was used to acquire experimental data for validation.

Main Results:

• The proposed method demonstrated robust defect detection performance on the experimental dataset.
• High classification accuracy was achieved, validating the effectiveness of the optimized SVM classifier.
• The integrated approach successfully identified and classified defects in fiber-optic coils.

Conclusions:

• The developed machine vision framework offers an optimal solution for the automatic detection of defects in fiber-optic coils.
• The combination of wavelet transform, nonlocal means filtering, and a genetically optimized SVM provides a highly accurate and efficient automated inspection system.
• This research advances automated quality control in the manufacturing of fiber-optic components.