A Traffic Monitoring System Based On Image Processing

Traffic monitoring system has recently received much attention due to the increase of traffic jams and the seriousness of accidents. Precise traffic data are the bases of right traffic control. How to acquire precise data of traffic information is the key step of intelligent traffic control. Traffic monitoring systems can rely on several detection methods to estimate traffic parameters in real-time. The computer video and image detection for traffic will replace many old methods of traffic detection due to its easy installation, lower cost, widely detecting scope and plentiful information.This thesis presents a real-time traffic monitoring and detecting system based on the principles of computer vision and digital image processing technique. By a video camera and a video-captured card, the video images of traffic information are transferred and digitized to digital images that could be processed by computer. Then, the qualities of images can be enhanced by filtering and noise reduction. By processing and analyzing image data of traffic information, we can acquire some data of traffic parameters for right traffic control, such as the vehicle speed and traffic flow.In the course of measuring the traffic parameters, a vision model between the road coordinate and image coordinate is needed. In this paper, we introduce the camera model calibration algorithms and adopt a method of moving a single camera to calibrate parameters of traffic monitoring system. Based on digital image differential detecting technology, we present a real-time method of vehicle detection. First, select background image and set the linear projection, and detect the moment when vehicle passes the linear projection, then compute the speed of vehicle and traffic flow. By using background-updating technique, we can detect vehicles more precisely. Traffic flow is an important traffic parameter, so traffic flow forecast is significant for traffic management, hi this paper, we apply neural network to research of traffic flow forecast and present a method of traffic flow forecast based on BP neural networkAt the end of this paper, an overview of the paper is presented and the existing problems are discussed.