Study Of Foreign Object Intrusion Intelligent Recognition On Railway Track

With the rapid development of railway in our country, the speed of train is continuously improved, railway operation security has become increasingly important. In recent years,China's railway line mileage increases gradually, the casualty accidents what the foreign object intrusion case are more and more, which seriously threaten people's life and property safety. Based on the study of the relevant literature at home and abroad related to the railway track foreign object intrusion and artificial intelligence, an intelligent recognition method of railway track foreign object intrusion based on digital image processing technology is researched, the system can accurately recognize foreign object and timely alert when foreign object intrusives railway track, it can provide technical support for railway safety operation. The main research contents are organized as follows:Firstly, the thesis deeply analyzes the facing and the existing problems of the railway operation safety supporting system, as well as the function and significance of intelligent identification foreign object for the train safe operation, in addition, based on the study of the existing railway security supporting system, the intelligent recognition project of the foreign object intrusion on railway track is designed.Secondly, the window of danger area foreign object detection on the railway track foreign object intrusion is determined in the thesis, according to the different shape characteristics of the straight railway track and curved railway track, the different methods for establishing their windows of danger area foreign object detection are designed. under the principle of follow the window of danger area foreign object detection in this thesis, for the straight railway track, the track edge extraction is completed mainly through the method of combining the improved Canny operator and Hough transform, and the track position is determined,basis on the track position, the window of danger area foreign object detection on the straight railway track is established; for the curved railway track, foreign object detection danger area is divided through the method of marking foreign object detection area in the actual scene, the window of danger area foreign object detection on the curved railway track is established through the method of dividing the image into many small pieces for processing and then combines them together.Thirdly, the method of railway track foreign object detection is studied. The basic principles of traditional background modeling methods is expouned, moreover, a double background modeling method is put forward in this thesis, the background difference method,frame difference method and optical flow method for image object detection are made a comparative analysis, the object detection method based on background difference method and frame difference method is designed to detect track foreign object, and the pseudo objectof the image detection results is eliminated. The simulation shows that the method can achieve good accuracy on the foreign object detection.Finally, the intelligent recognition method of railway track foreign object intrusion is studied, including two aspects: one is the method of foreign object recognition, it is completed through image correlation coefficient and the ratio of the foreign object area and the image area; the other is the method of the train recognition, which is the distinction between train and foreign object, the intelligent recognition of the train is completed according to the increasing or decreasing train area in the prospect of sequence images during train moving.The results show that the method has good effect on foreign object intelligent recognition and train intelligent recognition.Based on the theoretical analysis and simulation research, the conclusion of the thesis makes a summary for the intelligent recognition method on railway track foreign object intrusion, elaborates the thesis' s main research achievements, and put forwards some need further improvement and research work.