Application Research Of Railway Intelligent Monitoring Based On Moving Target Detection

In recent years,with the development of railway construction in our country,especially the rapid application of high speed rail to every city,the railway operation management is more and more intelligent,and the modern information technology is applied to the railway construction.The intelligent monitoring system is a direct,real-time and replayable monitoring and management mode.It is necessary to use some railway lines to protect railway operation and personal safety,to assist public security departments in combating crime and to maintain sustainable development of society.The technology of artificial intelligence deep learning has advanced the remarkable development of computer vision,and it is of practical value to establish an intelligent monitoring system based on the railway background environment.On the basis of studying railway intelligent monitoring,combining the actual background environment of Hainan East loop line high speed railway,combining the traditional moving target detection algorithm and artificial intelligence deep learning target detection algorithm,the application of Railway Intelligent Monitoring Based on mobile target detection is studied.The main work content is as follows:(1)according to the actual situation of the railway,the occurrence of other moving targets other than the train is prohibited within the railway fence,the alarm zone is set up along the railway line,and the video part of the alarm area is monitored and monitored in real time,including the pedestrian movement and the moving safety of the railway on the railway,and the railway duty people.The police can handle it in time and effectively.(2)preprocessing the video images collected from the actual railway scene.The video images collected from the field contain a lot of noise,mainly including Gauss noise and salt and pepper noise.Removing Gauss noise usually uses bilateral filters to remove salt and pepper noise,generally using median filtering.In this paper,an improved weighted bilateral filter is adopted to remove Gauss noise and salt and pepper noise at the same time.It is proved by experimental results.Better video surveillance images can be obtained.(3)an improved visual background extraction algorithm(VIBE)and a deformable component model DPM algorithm are used.According to the actual situation of the railway background environment,the high speed train causes the camera and the picture jitter by monitoring the picture very quickly.The high speed train leaves the shadow and hole after passing through,and adopts the adaptive parameter to weaken the shake.In order to eliminate the complex phenomenon,histogram matching is applied to eliminate the void phenomenon.Through the improved VIBE algorithm,the location area of the moving target is suspected,and the DPM algorithm is used to detect the target in the target location area.The experimental results show that the target detection effect can be better in the simple environment.(4)to design an improved Faster R-CNN algorithm for deep learning.In order to meet the real time requirement in the actual environment of high speed railway,the railway intelligent monitoring is studied on the basis of the deep learning convolution neural network CNN.The improved Faster R-CNN algorithm and the traditional improved VIBE algorithm are used to design a kind of improvement.The fusion of VIBE algorithm and Faster R-CNN algorithm,through the improved VIBE algorithm positioning to the existence of the mobile target area,and then in the region to get the Faster R-CNN target feature extraction,and finally the moving target classification and regression.In the end,a large number of experimental results show that the detection of illegal pedestrians and small animal moving targets in a variety of complex railway background environment shows that the algorithm can detect the moving target in real-time and effectively in the complex background environment of the high speed railway.