[IEEE 2019 Chinese Control And Decision Conference (CCDC) - Nanchang, China (2019.6.3-2019.6.5)] 2019 Chinese Control And Decision Conference (CCDC) - Research on Detection and Recognition Algorithm of Optical Fiber for Surface Defects

摘要： Based on machine vision technology, this paper studies the detection and identification technology of surface defects of optical fiber communication cables. This paper uses mathematical geometry and linear interpolation method to convert cable cylinder into plane. Based on the gray-scale characteristics of the pinhole in the picture, gray threshold and gradient amplitude are used to improve the detection rate of the pinhole. Finally, the experimental results show that this algorithm can all-around detect the surface defects of optical cable, with the minimum detection area of 0.05mm2, the diameter detection accuracy of 0.1mm, and real-time alarm for defects. In the case of a lot of dust, ensure the correct detection of conventional defects on the surface of the cable, the correct detection rate of the pinhole reaches 70%, reducing the false alarm rate of dust and water stains.