%0 Journal Article
%T Abnormal Behavior Detection Using Trajectory Analysis in Camera Sensor Networks
%A Yong Wang
%A Dianhong Wang
%A Fenxiong Chen
%J International Journal of Distributed Sensor Networks
%D 2014
%I Hindawi Publishing Corporation
%R 10.1155/2014/839045
%X Camera sensor networks have developed as a new technology for the wide-area video surveillance. In view of the limited power and computational capability of the camera nodes, the paper presents an abnormal behavior detection approach which is convenient and available for camera sensor networks. Trajectory analysis and anomaly modeling are carried out by single-node processing, whereas anomaly detection is performed by multinode voting. The main contributions of the proposed method are summarized as follows. First, target trajectories are reconstructed and represented as symbol sequences. Second, the sequences are taken into account using Markov model for building the transition probability matrix which can be used to automatically analyze abnormal behavior. Third, the final decision of anomaly detection is made through the majority voting of local results of individual camera nodes. Experimental results show that the proposed method can effectively estimate typical abnormal behaviors in real scenes. 1. Introduction Camera sensor networks consist of low-power microcamera nodes, which integrate the image sensor, embedded processor, and wireless transceiver. Multiple camera nodes with different views can provide comprehensive information and enhance the reliability of the captured events. Due to the advantages of enlarging surveillance area and solving target occlusion, camera sensor networks are best suited for real-time visual surveillance applications [1, 2]. One of essential purposes of visual surveillance is to detect moving targets and identify abnormal behaviors. In the past, model-based approaches have been proposed to tackle the anomaly detection problem. The work in [3] adopted dynamic Bayesian networks to model normal activities. An activity will be identified as abnormal if the likelihood of being generated by normal models is less than a threshold. Nevertheless an appropriate threshold is hard to determine in practice. In [4], a hidden Markov model (HMM) was applied to represent normal activities and perform anomaly detection. Note that it is difficult to label all the activities because of the tremendous variety of movement states. Trajectory modeling can determine the movement anomaly in video sequences, and many previous works have discussed the issue. In [5], vision-based trajectory learning and analysis methods were discussed. In [6], a sparse reconstruction analysis of target trajectory was introduced to detect abnormal behaviors. However, most of these works employ supervised learning to recognize normal behaviors, which requires a
%U http://www.hindawi.com/journals/ijdsn/2014/839045/