Research On Intelligent Identification And Automatic Forewarning Of Foreign Objects Intrusion In Rail

The technical development of high-speed railway is changing with each passing day,and the train running speed has also been greatly improved.This greatly facilitates people’s travel,but with it comes the consideration of train safety.The study found that a major cause of train accidents is foreign objects invasion incident,which not only greatly endangered people’s lives and property,but also caused inestimable loss of state property.On the basis of studying the foreign objects intrusion monitoring technology of rail transit at home and abroad,this thesis focuses on the monitoring method of foreign objects intrusion limitation based on video image processing,and realizes the real-time monitoring of foreign objects intrusion limitation along the railway,to a certain extent ensure the safe operation of the train.The main research contents of the thesis are as follows:Firstly,the method of establishing a dangerous area and the method of optimizing the detection environment are studied.For straight railway track,the Hough transform is used to extract rail track edges features;For curved railway track,segment the rails first to obtain segmented straight rails,then use Hough to extract the rail track edges features,and then combine them to obtain the curved track edge extraction.According to the actual requirements for the establishment of dangerous areas,the rails are used as the basis to establish four different types of railway track dangerous areas that are suitable for straight lines,curves lines,multi-track straight lines,and multi-track curves lines.By analyzing shadow features and lighting models,a static shadow automatic detection and removal algorithm based on Support Vector Machine(SVM)and region sub-block matching is proposed to optimize the detection environment.Secondly,the object detection algorithm is studied.Aiming at the defect of Visual Background Extractor(ViBe)algorithm in real-time monitoring and low warning accuracy due to slow ghost elimination in detecting dynamic invading foreign bodies on railway tracks.This paper uses background reconstruction and pixel block replacement method to quickly suppress ghosts in different periods;Aiming at the problems of low detection accuracy and poor anti-interference ability for dynamic intrusion of foreign objects in complex rail environment,a real-time detection method for foreign object intrusion in railway track based on improved Mixture of Gaussian-Low Rank Matrix Factorization(MOG-LRMF)algorithm is put forward.After simulation experiments in different scenarios,both improved algorithms can achieve higher detection accuracy.Finally,the intrusion foreign objects identification and automatic warning methods are studied.Intelligent recognition of dangerous foreign objects by using the ratio of pixels of dangerous foreign objects to image pixels;According to the running state of the train in different periods of time,by analyzing and counting the number of pixels occupied by the train in the detection results of moving objects in 3 consecutive frames and the changing trend,the difference between the train and the encroaching-pedestrian is made,and the train is recognized.Simulation experiments show that the method can accurately identify and distinguish dangerous foreign objects and trains.