一种基于视频的车站人群密度估计算法

提出了一种基于视频的车站人群密度检测算法。该算法能标记出无人活动区域和有人活动区域,在检测人群密度方面只计算有人活动区域,可降低计算复杂度,为人群密度检测的实时性奠定了基础;在计算人流密度时,考虑到了摄像头画面出现黑屏、雪花和移位的现象,从整体上提高了人群密度检测的准确率;通过对有人运动区域的连通域内像素数与有人区域的像素点数的比值计算人群整体密度值,计算复杂度低且衡量人群整体密度的精确度较高。此方法尤其适用于机场、地铁、车站等场景模型下人群密度的实时场景监测。

Estimate of crowd density at the station based on video monitoring

Crowd density estimate is important for public safety and timely management. An algorithm of crowd density estimate at the station based on video surveillance is presented in this paper. The algorithm can distinguish inactive and active areas, and only estimate crowd density of active areas so that it reduces greatly computation complexity. When estimating crowd density, it takes into account the black screen, snowflakes and shift, which may appear in the camera pictures. As a result, it can improve the accuracy of estimation. The ratio of the number of pixels of activity area to whole area is the estimated value of the crowd density. The algorithm can not only lower computation complexity but also enhance the accuracy. This method is particularly suitable for real time monitor of crowd density, such as in airports, the Metro and railway stations.