一种基于视频能量的异常事件检测算法研究

An Approach of Abnormally Detection based on Crowd Energy

监控人群拥挤区域的行人流动情况,对于各类安检任务有着重要作用。该问题通常涉及两个内容:对拥挤的程度进行评价及判断其异常与否,这需要对大量运动信息进行研究;定义了一种群体能量来进行实时监控系统的拥挤人群建模与处理,通过对能量曲线进行分析,提出了一种检测群体异常事件的方法,从实验结果可见该方法的有效性。

Surveillance on the flows of crowded pedestrian area has an important role.The problem usually involves two elements： to assess the crowdedness and to determine whether it is abnomal.This requires a lot of motion information.Here we denes a crowd energy to carry out real-time monitoring system of crowd modeling and processing.Through analysis of the energy curve,a method of monitoring an abnormal event of the scene is presented.The result showing that the method is effective.