| With the rapid development of China's economy,rail transportation has become an important means for more and more cities to solve urban traffic congestion.The rail transportation security problems that accompany it are also more prominent.But the application of video processing technology to rail transportation safety is an important practice,which involves frontier fields such as machine learning,pattern recognition,and distributed technology.Video detection methods for dangerous phenomena and dangerous behaviors affecting train safety in rail transportation are mainly provided in this paper,such as the detection method of electric flashing arc,the discrimination method of trains leaving normal track,and the system framework of rail transportation video monitoring of simultaneous and real time.In this paper,a background selection method based on time-average is provided,and a background update method based on re-selection of background and weighted background update is given.It can adaptively adjust the background according to the change of video picture;improved erosion and expansion algorithms based on erosion and dilation algorithm of morphology are given to obtain better foreground target optimization effects;based on the HSV color space,a de-shading method suitable for eliminating the shadows of strong sunlight on the following cars is provided,and the adverse effect of shadows on train target recognition is avoided.The concept of dispersion is applied to image processing to reduce the false alarm rate for identifying dangerous phenomena.A video detection method for the dangerous phenomenon of electric flashing arc is given in this article.Based on the rule and duration of gray image of electric flashing arc,a target tracking electric flashing arc detection method is provided to separate the common target from the arc;based on the law of jitter of the electric flashing arc at discrete pixels,a method for detecting the electric flashing arc in the pixel point dimension is given,which can detect the flashing arc phenomenon in more complicated scenes.Based on the background of the image and the saturation law of the electric flashing arc,an adaptive background selection method is given.The video detection method of trains out of normal track risk behavior was studied.Based on the method of target recognition in computer vision,the discrimination method of the train leaving the normal track is given.According to the gray features of the image,a dual threshold method for day and night is given,and the difference detection is performed under different scenarios.The basic framework of the rail transportation safety video detection system is designed and implemented in this paper.Based on the MFC framework of the software platform,multi-threading,process programming model and event triggering model are adopted to achieve real-time acquisition of multiple channels of high-definition video data,decoding of video streams,detection of dangerous events,capture of alarm evidence,real-time transmission of alarm messages and humanized client display.This system is applied to the rail transportation system,which can accurately and effectively monitor the rail train,find the main dangerous phenomenon when the train is running and send an alarm message in time,and record the incident at the time of the video and pictures to allow staff to view.This system greatly increase the safety of rail transportation. |
PDF Full Text Request