Research On Non-Contact Detection System Of Rail Wear

Railway is the major artery of the national economy,one of the most important modes of transport,and undertakes major transportation tasks.The rapid development of China’s economy has led the development of rail transport to high-speed,high-density and heavy-load.With the increase of train speed and traffic density,the wear of rail is becoming more and more serious.How to detect rail wear more quickly and more effectively is one of the main problems in rail detection.Currently,China is mainly dependent on human inspector,which method is poor efficiency and accuracy.Because the detection system based on machine vision has the advantages of structure,efficiency,precision and automation,the rail detection based on machine vision has become one of the hottest areas of rail detection in recent years.So in this paper,a set of simple and portable rail wear detection system is designed with the technology of machine vision and image processing.In this paper,through the analysis and comparison of various calibration methods,the calibration method by Zhang Z.Y is chosen,combined with the characteristics of rail wear detection to determine the calibration program to achieve camera calibration.The process of the rail contour processing algorithm is designed,and a matching method combining feature point matching and ICP algorithm is proposed,which is divided into matching and fine matching to achieve more accurate matching of rail and turnout.This paper firstly introduces method and development of rail wear detection,and puts forward a non-contact rail wear detection program based on computer vision.The overall design of the system and hardware is developed based on the basic theory of machine vision technology.Through the research and analysis of the camera calibration method,the calibration program is designed and the feasibility of the program is verified by experiments.Considering the hardware,the algorithm of image processing is designed,including noise filtering,region of interest extraction,color segmentation,centerline extraction,contour recovery and contour matching.With the help of OpenCV algorithms library,the system is realized on the VS2013 platform.The function of image collection,data processing,inquiry and report generation also achieved.Finally,the experiment is carried out in the laboratory using the rail wear detection system.The experiment suggested that this system can accurately measure the rail wear,and the accuracy meets the engineering requirements.It has good application value and practical significance.