中文摘要：

光纤传像元件缺陷检测一般都是由人工来完成的,其效率低下、误差大并且成本高.本文设计完成了一套基于机器视觉的光纤传像元件缺陷检测系统.根据对于光纤传像元件的有关知识了解,本系统选用双远心工业镜头配合千万像素级大靶面的CCD工业相机,实现了光纤传像元件表面缺陷的一次性成像并保证了足够的分辨率.在图像分割检测算法上提出了一种基于改进FCM的算法,通过相关实验验证并和传统的图像分割算法——全局阈值分割法做比较得出：该算法能够提升缺陷检测效率和准确率,进而能够准确而有效地实现缺陷的分割.整个系统运行稳定可靠,可以满足检测需求.

英文摘要：

Defect detection of optical fiber devices for image transmission was generally completed by manual work. It had low efficiency, big error and high cost etc. The article has designed and accom- plished a set of defect detection system of optical fiber devices for image transmission based on machine vision. According to the understanding of relevant knowledge for optical fiber devices for image trans- mission, the system chooses the pair of telecentric industry lens with tens of millions of pixels of large target surface CCD industrial camera, to achieve a onetime imaging on optical fiber devices for image transmission surface defects and ensure suffieient resolution. In the detection algorithm of image seg- mentation, an algorithm which is based on the improved FCM is proposed. By means of the relevant ex- perimental verification and comparing with the traditional image segmentation algorithm-global threshold segmentation, the algorithm can promote efficiency and accuracy rate of defect detection, and then can realize segmentation of defect accurately and effectively. The whole of system operates stably and relia- bly, and ean meet the needs of detection.