Automatic defect detection of carbon fiber woven fabrics using machine vision

Abstract

In the context of defect detection during the preform preparation stage in carbon fiber composite liquid molding processes, this study proposed three machine vision-based automatic detection methods, i.e. Gabor filter-assisted detection (GFAD), morphological processing-assisted detection (MPAD), and integral image-assisted detection (IIAD). These methods were evaluated on carbon fiber fabric images containing nine different types of defects, considering four evaluation aspects. The experimental results demonstrated the effectiveness of these three methods in defect detection. Among them, the best-performing method achieved a detection accuracy of over 90% for defect sizes and a recognition rate of 98% for defects.