Traffic Surveillance System for Vehicle Flow Detection

Abstract

Robust and reliable traffic surveillance system is an urgent need to improve traffic control and management. Vehicle flow detection appears to be an important part in surveillance system. The traffic flow shows the traffic state in fixed time interval and helps to manage and control especially when there's a traffic jam. In this paper, we propose a traffic surveillance system for vehicle detection and counting. First, we extract ROI (region of interest) from the traffic video using the combination of background subtraction and edge detection. Second, after detection of the traffic, an automatic lane-dividing method is presented to acquire several lanes. At last, based on the width of lanes, we set several self-adapting windows for vehicle counting instead of the traditional fixed window method. The self-adapting windows can eliminate the effect of vehicle sight distance on road because their width are adjusted by the width of lanes. Experiments in real traffic video have proved the robust and effectiveness of our system.