Design Of Video Surveillance System Algorithms For Railway Crossing Based On Image Processing

With the development of the railway, railway transport rates are also rising, but because of many railway crossing lack of management and people have a weak awareness of road safety and many other factors, therefore, railway traffic accidents menace people's live, properties and country's transportation. The safety of railway crossing has always been very concerned about the various countries.Research the railway crossing safety device is particularly important. At present, there are many types of railway crossing safety devices, and they work of different mechanisms. In this paper, we chose the railway crossing video surveillance system. The working principle is to trace, recognize, and classify the railway crossing by surveillance the rail field. If the fault is detected, the system alarm whistle at the same time, then to send the data as some required codes to the train's special receiver through the wireless transport devices, and the train driver can judge the situation of the ahead railway crossing according to the received codes and take some measures to determent the incoming accident at the same time. So the accident rate could be decreased, the railway transportation safty will be enhanced.This paper analyzed the system functional object and performance required, in order to determine the core of the system is confirmed to be image processing module. The main contents of the thesis include system algorithm design, algorithm validating. The algorithms can be divided into five parts: pre-image, object detection, object tracking, object recognizing and fault detection. Pre-image is the basic and important part, the main work is to improve the original image, to prepare for the image analysis; object detection is the prerequisite of object tracking and object recognizing, the main task is to extract out the static or movement objects from stationary background; object tracking provide a basic for the high-level image analysis, this technology is to forcast the movement trend of moving targets; object recognizing provid a basic for fault detection; the final outcome is fault detection, and we are most concerned about the conclusions, then all the algorithms have been validated in MATLAB.