Automatic Detection System Of Optical Fiber End Surface Defects Based On Machine Vision

As a kind of optical flexible transmission medium,optical fiber has been widely used in laser processing,optical communication and biomedical fields,and the defects of fiber end face will have a serious impact on the fiber itself and even the optical system.Therefore,the detection of the fiber end face is an indispensable part before the fiber works or after it is polished.Optical fibers have different acceptance criteria according to their types and fields.Nowadays,the most mainstream detection method is still manual detection,but this method is subjective and inefficient,and its accuracy is not high.With the rise of artificial intelligence and the proposal of "Made in China 2025" plan,machine vision technology has been rapidly developed and applied to various fields,making the products detection more and more automatic and intelligent.In order to improve the deficiency of existing fiber end face detection technology,this paper mainly studies the automatic detection system of fiber end face defects based on machine vision.Firstly,taking the acceptance criteria of fiber end face as the starting point,this paper presents the overall technical scheme of the system,and the detection process is divided into several processes,such as image collection,image processing,analysis &judgment,this paper also designs the software architecture of the system.Secondly,the algorithm of each step of different processes in image processing is studied,then a set of the most suitable algorithm is selected and integrated,with this set of algorithm,different types of defects can be detected simultaneously.At the same time,a general composite algorithm is presented to automatically process the fiber end face image.The algorithm and related parameters are applicable to different areas of the fiber end face.For the same type of fiber end face image,no algorithm and parameters needs to be changed,so that the defects detection enables the system to be efficient and automated.The system mainly focuses on the computer,using the fiber end face detector to complete the collection and transmission of the fiber end face image.The software part mainly improves the image processing algorithm through Java language,using the OpenCV open source algorithm library to complete the system construction.And an intuitive graphical user interface is established for better human-computer interaction and for the outputs.The system can achieve the functions of fiber positioning,defects detection by different classifications,and it can also quantify the defects detection results.The result of different fibers' detection shows that the system can accurately check before the acceptance of the fiber end face,this system can achieve the expected effect.