Research On Railway Fastener Defects Method Based On Image Processing

As a necessary connection between rails and sleepers,the railway fastener is an important component for maintaining the safety of railway transportation.It is mainly used to fix the correct position of the rail,prevent the rail from being displaced in the longitudinal and lateral directions,provide an appropriate amount of elasticity and transfer the force of the rails to the sleeper or track bed rail.The lack of fasteners makes the rails not completely fixed,which seriously affects the safe operation of the train.In severe cases,it may cause major accidents such as derailment of trains.With the new demand for track fastener detection under the rapid development of modern railways,the traditional manual detection has poor real-time and accuracy in actual detection.Therefore,how to realize the detection and detection of railway fastener status quickly and accurately is very important.The thesis combines a railway fastener image acquisition system with mature image processing technology to realize railway fastener identification and detection.The main work of the thesis is as follows:Firstly,according to the basic requirements of image acquisition of railway fasteners,analyzing the actual situation of image acquisition on site,camera imaging principle,and the characteristics of light source,selecting the hardware equipment of the specific acquisition system.At the same time,according to the principle of fastener defect detection,the algorithm flow of this thesis is introduced.Secondly,the image will generate noise during the acquisition process,and the railway fastener image needs to be filtered and denoised.The median filtering can remove the false edges caused by the salt and pepper noise,while the guided filtering algorithm has better edge retention characteristics.Therefore,the thesis adopts an improved guidance filtering algorithm to denoise railway fastener images,which is beneficial to retaining the edge features of the fasteners and ensures the accuracy of fastener positioning and recognition in subsequent railway fastener images.Thirdly,analyzing The Canny edge detection algorithm and the widely used Opencv,combining the traditional Canny algorithm with the adaptive function in Matlab,this thesis use the adaptive Opencv-based Canny algorithm to detect the edge of the railway fastener image.According to the result of edge detection,the position of the rails and sleepers in the image is located by the gray integral projection algorithm,and the fastener area is roughly positioned.Then the template matching algorithm is used to achieve the precise positioning of the fastener area which is based on the roughly positioning.Finally,respectively extracting the Local Binary Pattern(LBP)features of the fastener edge features and the Pyramid Histogram of Oriented Gradient(PHOG)features of the fastener texture features.In order to obtain a feature vector that more accurately characterizesof the fastener feature,the above two types of features are merged.Then,the fused feature vector is input into the Support Vector Machine(SVM)to realize the state identification of the railway fastener.The simulation experiment proves that the fastener detection system of this thesis improves the accuracy of railway fastener defect identification and provides a reliable basis for railway patrol workers.