虚拟窗口阴影去除算法的车流量检测研究

Reserch of shadow removal algorithm for traffic flow detection based on virtual window

针对城市道路交通流量检测中的实时性和准确性的要求,在背景差分的基础上提出了一种改进的基于虚拟窗口检测的方法。通过获取前景目标尽量少的帧图像,快速建立虚拟窗口的初始背景模型,并实时更新背景;将Sobel边缘检测算法引入前景目标检测,从而快速检测出前景目标变化的完整区域,提高检测的准确率;再使用基于HSV色彩空间直方图势函数去除阴影算法,进一步去除前景目标中的阴影区域,有效地保留了真实的运动目标区域;最后统计交通车流量,可结合其他信息(如红绿灯状态)做出该路段相应的交通流状况判断。通过实验结果证明,所提出的检测算法可有效应用于视频交通车流量检测中。

In view of the accuracy and real-time requirements of urban road traffic flow detection, a new method based on virtual window detector is proposed. The initial background model of the virtual window is established quickly by acquiring the frame images of target prospects as little as possible, and the background is updated in real-time. In order to quickly detect changes in the entire region of foreground objects, the Sobel edge detection algorithm is introduced to the foreground object detection and improve the accuracy of detection. In order to further remove the shadow of the foreground object region and effectively retain the true movement of the target area, the potential function based on the HSV color space is used to remove the shadow histogram algorithm. Finally, traffic flow statistics can be combined with other information （such as traffic light status） to judge the road traffic flow conditions. Experimental results show that the proposed detection algorithm can be effectively applied to video traffic traffic flow detection.