一种基于视频的车辆检测算法

本文讨论了对传统帧差法的改进，并结合边缘检测法，提出一种环境自适应能力强、计算量小，适合于运动和静止车辆同时检测的车辆检测新方法，可正确判断有无车辆、完成车辆的计数，进而实现车流量计算、车速估计。在计数算法中采用预估校正、相关性修正等措施，提高了检测精度，大大改善了车辆检测效果，从而为交通监控系统提供实时有效的交通参数。

The New Vehicle Detection Algorithm Based on Video

This paper discusses the improvement of Frame-Difference method. Furthermore,combining Edge-Detec- tion metiod,it puts up a new vehicle detection method which adapts to the environment well and calculates very little. With this new method,moving and still vehicles can be detected simultaneously. So that Vehicle counting,speed esti- mating,and the information whether there are vehicles or not are achieved. In this algorithm,pre-estimate correction and relativity correction are successfully used to increase detection precision and to improve the detection effect. Con- sequently,real-time and available traffic parameters can be obtained for the traffic monitoring system.