基于人流量检测的改进CN算法

Improved CN algorithm based on pedestrian flow detection

为实现在行人严重遮挡时人流量的精确统计,研究一种基于人流量检测的改进CN算法。结合背景差分与三帧差分提取运动目标前景;通过梯度方向直方图与支持向量机判断头肩特征;在Kalman滤波器预测下一帧图像中目标位置的周围选取检测窗口,利用融合HOG与CN(颜色名)特征的改进CN算法实现目标跟踪;以感兴趣区域计数线为准,结合目标运动轨迹实现人流量统计。实验结果表明,该算法在有行人严重遮挡的情况下具有较高的检测效率。

To achieve accurate statistics of pedestrian flow when pedestrians are severely occluded,an improved CN algorithm based on pedestrian flow detection was researched.The moving target foreground was extracted by the combination of background difference and three-frame difference.The head and shoulder characteristics were judged by the histogram of oriented gradient and the support vector machine.The detection window was selected around the target position in the next frame image predicted using the Kalman filter,and the target tracking was implemented using the improved CN tracking algorithm fused HOG feature and CN(colour name)feature.The pedestrian flow was calculated by combining the target motion trajectory with the counting line of the region of interest.Experimental results show that the proposed algorithm has high detection efficiency in the case of severe occlusion of pedestrians.