Comparative Study on Trajectory Outlier Detection Algorithms

This paper explores, reviews, and evaluates the existing outlier trajectory detection using small, large and big databases. We divide existing solutions into two main categories: similarity-based, and clustering-based approaches. The first category groups solutions employing distance and neighborhoods computation to derive local density estimates. The second category explores the correlation between the trajectory data by using clustering techniques. Solutions in each of these categories are explained, implemented and compared using different trajectory database sizes. Compared to state-of-the-art survey papers, the contribution of this paper is in providing a deep analysis of all kinds of trajectory outlier detection methods. In this context, we can better understand the intuition, limitations, and benefits of the existing outlier trajectory detection algorithms. As a result, practitioners can receive some guidance for selecting the most suitable methods for their particular case.