基于整体特征的人群聚集和奔跑行为检测

Crowd gathering and running behavior detection based on overall features

建立高效完备的智能视频监控系统已经成为当今社会的迫切需求。针对公共场所人群聚集和奔跑的两类异常行为,提出一种基于分布熵和平均运动速度的检测方法。一方面,根据前景图像的空间分布情况,采用分布熵衡量场景中人群的集中程度,实现对场景中人群聚集行为的检测;另一方面,检测图像上的角点,采用光流法对这些角点进行跟踪并提取出产生运动的角点,进而获得运动角点在视频序列中连续两帧间的运动向量,计算出人群整体的运动速度,从而检测人群的奔跑行为。所提出的方法不需要对单个行人进行分割以及样本的训练。采用不同场景和不同人群密度下的视频对所提出算法进行验证的结果表明,本文方法可以快速、准确地进行人群异常行为检测。

Efficient complete intelligent video surveillance system has become the urgent needs in today′s society.For crowd gathering and running behavior detection in public places,a method based on distributive entropy and average speed of crowd movement is proposed.On the one hand,according to the spatial distribution of the foreground pixels,the distribution entropy is defined and used to measure the concentration of the crowd and implement the detection of the gathered crowd.On the other hand,the motion vector of the feature points between two consecutive frames in the video sequence is gained by tracking the feature points on the image using the optical flow method,which is used to estimate the crowd′s speed and detect the crowd running behavior.This method does not need to a single pedestrian segmentation as well as the training sample.To test the performance of this algorithm,videos with different scenes and different crowed densities are used in the experiments.Experimental results show that our method can quickly and accurately detect crowd abnormal behaviors.