Towards Online Anomalous Trajectory Detection and Classification

Anomalous trajectory detection has become an important research problem and received much attention in recent years. A few studies focus on identifying abnormal trajectory patterns and present feasible algorithms for anomalous trajectory detection and classification (ATDC). However, these algorithms fail to classifying patterns for a traveling trajectory, limiting their application in real-world scenarios. This paper proposes a fast algorithm for online anomalous trajectory detection and classification named FastOATDC. FastOATDC is an improved version of the previous algorithm FastATDC, which compares an ongoing trajectory with similar sequences of historical trajectories, leading to online detection and classification. Extensive experiments on real datasets have shown that the classification accuracy achieved by FastOATDC using only half of the trajectory can almost meet or even exceed the accuracy achieved by using the whole trajectory.