摘要
视频监控在社会安全方面扮演着越来越重要的角色,在计算机视觉领域,人群异常行为检测也成为非常重要的研究课题。提出一种基于运动熵计算的人群异常检测方法。该方法在图像上散布特征点,运用光流法分组计算出各特征点的运动大小与方向,并据此建立运动直方图,用图像熵的计算方法得出图组的运动熵,运动熵及平均能量值则作为异常检测的判断依据。实验使用明尼苏达大学逃离与恐慌相关实验视频及部分网络视频,实验表明此方法拥有较强的容错能力,并能实时正确的检测出大部分异常行为的发生。
Video surveillance in crowded areas is becoming more and more important for public security, abnormal crowd behavior detection is an important research issue in computer vision. Presents a method for the detection of abnormality in crowded scenes base on the Moving Entropy method. A grid of particles is placed over the image, the directions and sizes of these particles are based on optical flow techniques, a moving histogram is established for the moving entropy, the moving entropy and average energy is used to be the parameter for abnormal detection. The experiment are conducted on a publicly available dataset from University of Minnesota for escape panic scenarios and some crowd videos from the Web. The experiments show that this method has good fault tolerance and could correctly detect most of the abnormal behaviors in real time.
出处
《现代计算机》
2013年第5期40-43,共4页
Modern Computer
关键词
视频监控
人群
异常检测
运动熵
Video Surveillance
Crowd
Abnormal Detection
Moving Entropy
