Fiber optic connector end-face defect detection based on machine vision

Abstract

As an important signal connector in communication data transmission, the performance of optical fiber is closely related to the reliability of data transmission. Currently, most manufacturers still use manual visual observation under a traditional microscope for fiber end-face defect detection, which suffers from low precision, low efficiency, and poor consistency. This study provides a machine vision-based method for identifying defects in fiber optic connector end face called the POL detection method. The method can be used to detect defects such as oil, dust, impurities, dents, and scratches on fiber optic end faces. The experimental results show that the POL detection method has a combined detection accuracy of at least 97.14%. The present technology indicates specific methods for realizing image acquisition, image preprocessing, feature extraction, and defect localization. Compared with the existing traditional manual visual inspection, the accuracy of the method is improved by about 20% and the efficiency is improved by about 6 to 7 times.