Research On Rail Surface Defect Detection And Location Based On Image Processing

With the rapid development of high-speed railway,ensuring the safe running of trains has become the first priority of railway development.However,under the influence of high density,heavy haul trains and external environment,there are more and more defects on rail surface.If the rail surface defects are not dealt with in time,the expansion of these defects will have an impact on the safety of railway transportation,and even a rollover accident will happen in severe cases.Therefore,it is necessary to detect the rails surface defects timely and accurately,and transmit the defect geographical position and the related information of defects to the server side,and remind the maintenance personnel to handle in time.On the basis of the existing defect detection for rail surface,the detection algorithm of rail surface defect and the location of the defect geographical position are studied in the thesis.The main research contents are as follows:Firstly,taking the testing requirements of rail inspection image as the standard,the light source,camera and lens of the image acquisition device are selected;and selecting the location module which equipped with the track inspection car by comparing the existing train location method.Secondly,a maximum luminance sum algorithm based on RGB color model is proposed according to the luminance difference between rail surface area and the non-rail surface area,and realized the extraction of rail surface area;the median filter is used to de-noise,and the gray contrast image algorithm is used to enhance the image after filtering.Thirdly,through the analysis of existing rail surface defect segmentation methods,combined with the advantages of morphological reconstruction and the maximum entropy threshold method can be well segmented for simple images,a rail surface defect segmentation method based on the background difference and the maximum entropy threshold method is proposed;according to the analysis of the connected domain,the pixels of all the defects are counted,and find out defects area that are larger than 25 square millimeters,then marked them with the minimum bounding rectangle,and extracted the defect at the same time.Finally,if the rail image contains defects larger than 25 square millimeters,the geographical position of the defective picture will be located,the geographic position and related information will be transmitted.The SIM868 main module is configured in the first place,which integrated with the GPS(Global Positioning System)module and the GPRS(General Packet Radio Service)module,and TCP connection was established;then the simulation test on whether GPS module can locate correctly and GPRS module can transmit timely is carried out.The overall simulation of the system shows that the defect detection method studied in thesis can be used to detect the rail surface defects,GPS module can locate the geographical position of defect,GPRS module can accurately send the packets which contains the location information obtained by GPS and related information containing defect images to the server side.