Algorithm On Locomotive Intelligent Video Monitoring System

Locomotive Intelligent Video Monitoring System is a system which insures the train's travelling safety. By using this system, road safety has been improved. Simultaneously, the mode of the train's security has been transformed from manual-control mode to machine-control mode, and finally realize total machine-control mode.The Intelligent Video Monitoring of this issue is forefront topic which is under the spotlight and a new application direction. It's the crystallization of many high-tech subjects, such as, Machine Vision, Image Engineering, Pattern Recognition, and Artificial Intelligence.The Intelligent Video Monitoring automatic analyses Video-Image sequence which is recorded by camera by using video analysis method and Computer Vision. The order of this analysis is to realize the location and identification of target in monitoring scenes, and on this basis, to analyze and judge the monitoring target's behavior in order to get the comprehension of meanings of image content and explanation of the objective scene, then, guide and plan the action.The whole work of this issue is studying three parts of theories and methods which are belonged to Intelligent Video Monitoring; they are the location and identification of human targets, the identification of locomotive equipments, and the detection of railway blocks. This study is based on many theories of Intelligent Video Monitoring.Throughout the studying process of the identification of human targets and locomotive equipment, this issue has designed three dynamic scene update methods including Regional Update method, Variance Update method, and Edge Update method, and two foreground-extract methods which include Background Image Difference method and Background Subtraction method to adapt the complex and volatile external environment. Meanwhile this issue realizes real-time alarm function. In research process of the detection of railway blocks, this issue extracts the line information of rails by using Hough transform algorithm, in order to locate the blocks and realize forecast function of scene information. In the whole design process of intelligent algorithm, this issue uses a lot of Digital Image Processing methods, such as Grey-Level Transformation, Edge Detection, Image Segmentation, and Threshold-Image Projection Theory to realize the function of computer intelligent identification.All the methods designed by this issue provide some certain reference value for the study of Intelligent Video Monitoring algorithms.