Research On Surface Defect Detection Technology Of High-speed Rail Based On Machine Vision

With the rapid development of high-speed railway technology,its fast,punctual and comfortable features have brought great convenience to people's travel.China's high-speed rail has become a beautiful business card for national diplomatic cooperation.In order to ensure efficient and safe operation of high-speed railways,it is necessary to quickly,real-time and automatically detect the health status of the rails.Aiming at the problems of low efficiency,poor precision and potential safety hazards caused by traditional manual visual inspection and non-destructive testing,this paper designs a set of high-speed rail surface defect detection system by machine vision detection method,which replaces human eyes with machine and realizes track surface defects.Online real-time intelligent detection.Firstly,the paper introduces the research background and practical significance of high-speed rail surface defect detection,expounds the types and causes of rail surface defects,and proposes the application of machine vision detection technology to rail surface defect detection in view of the shortcomings of common detection techniques;Secondly,the main functions and related requirements of the rail surface defect detection system are analyzed.Based on this,the overall design scheme of the system is proposed,and the types of light source,lighting scheme,industrial camera and lens are introduced in detail,which are selected according to the system related performance indicators.The hardware parameter type has built a set of hardware devices with simple structure,adjustable light source and convenient operation for high-speed rail surface defect detection system;Then,aiming at the difficult problem in the detection of track surface defects,a tracking surface defect detection algorithm based on Blob analysis is proposed and implemented.In order to highlight the defect characteristics of the rail surface and reduce the time of subsequent defect detection,an adaptive gain contrast enhancement algorithm and an adaptive vertical projection method are studied to accurately extract the ROI region of the track surface.By comparing OTSU,maximum entropy and iteration three threshold segmentation algorithms,the selected iterative threshold segmentation method accurately and quickly segmented the surface defects of the rails,and obtained the real defect regions according to the actual detection requirements.At the same time,the seed filling algorithm was studied to solve the defects.Analysis and marking.The experimental results show that the algorithm can detect the main defects of the rail surface quickly and accurately.Furthermore,the feature extraction and analysis of scar and corrugation scratch information were carried out.A SVM-based high-speed rail surface defect recognition algorithm was studied and implemented.The most effective feature is selected as the input feature vector of the classification model.At the same time,the RBF is used as the kernel and the network search method is used to optimize the parameters,and the SVM classification model is trained.Based on the classification model,the surface defects of the rail are identified.The experimental results verify that the SVM classification model can effectively and accurately identify the two surface defects such as scar and corrugation.Finally,the high-speed rail surface visual defect detection and recognition software was designed and developed,which effectively realized the control of the hardware part and the real-time monitoring and identification of the surface defects of the rail.