期刊名称：IAENG International Journal of Computer Science
印刷版ISSN：1819-656X
电子版ISSN：1819-9224
出版年度：2019
卷号：46
期号：2
页码：311-320
出版社：IAENG - International Association of Engineers
摘要：It is a crucial issue to efficiently detect anomaly from surveillance videos. Abnormal behavior detection is developed in an unsupervised way based on spatio-temporal motion analysis in pedestrian virtual space. The pedestrian virtual plane is constructed which consists of both the ground plane and the pedestrian head one. The abnormal behavior is discriminated by a circular variance of pedestrian trajectories around the 3D virtual region instead of traditional 2D protected one. The protected region can be assigned as different shapes and sizes. Experiments show that the proposed method is efficient for distinguishing the anomaly in a protected region without any hypothesis for the scenario contents in advance. Comparisons with state-of-the-arts highlight the superior performance of the proposed method.
