Research On Optical Fiber Connector Face Inspection

Fiber optical is a transmission medium for optical communication with decades of developing history as the most significant communication tool in the modern society. In the practical application, the coupling between different sections of fibers is accomplished through optical fiber connector. The end face inspection for optical fiber connector is therefore an indispensable process in optical communication since the transmission reliability and efficiency are largely depended on the geometric morphology, cleanliness of the optical fiber end face.In order to meet the needs of practical application, this paper studies the image processing algorithms on optical fiber connector end face inspection and realizes measurement of cladding diameter, packet layer center, packet layer non circularity geometric parameters of optical fiber in the optical fiber connector under the complex background. The main contents and achievements are as follows:1. This thesis described the current status of optical fiber connector end face inspection methods. Two kinds of common optical fiber connector end face inspection methods, the interference measurement method and microscopic imaging method, were analyzed and summarized.2. Based on microscopic imaging visual detection principle, the common filtering and edge detection algorithm were studied. The application of image processing algorithm in optical fiber connector interface detection project was realized, and the image preprocessing and fiber core edge detection were completed. These algorithms were proved by simulation and experimental analysis as an important operation of the fiber segmentation and positioning for latter processing.3. Regarding the complicated background noise of the optical fiber interface, we proposed a profile detection algorithm combining the rough positioning and fine positioning. In this method, first of all, we accomplished coarse positioning and segmentation for single optical fiber connector interface using the gradient Hough method, then extracted the correct optical edge in the segmented area with the help of curvature features. Finally we realized fine positioning by fitting points on these the noise-excluded correct edges with least squares method. The experimental result showed that the proposed method in the combination of coarse and fine positioning is more accurate and robust.