Title :
Ship movement anomaly detection using specialized distance measures
Author :
Bo Liu;Erico N. de Souza;Cassey Hilliard;Stan Matwin
Author_Institution :
Faculty of Computer Science, Dalhousie University, Canada
fDate :
7/1/2015 12:00:00 AM
Abstract :
This paper provides a solution for anomaly detection in maritime traffic domain based on the clustering results presented in a previous work. That work created clusters for vessels moving close to shores by associating vessel movements with International Maritime Organization Rules (especially Traffic Separation Scheme Boundaries). In this paper, we show how three division distances with the clusters can detect anomalous navigational behaviors. The proposed method decides for each trajectory point if the vessel is anomalous, considering longitude, latitude, speed and direction. Although the approach is point-based, which is applicable for real-time AIS surveillance, it is also flexible enough for analysts to set their own threshold for labeling whole trajectories.
Keywords :
"Trajectory","Data models","Gravity","Clustering algorithms","Data mining","Marine vehicles","Labeling"
Conference_Titel :
Information Fusion (Fusion), 2015 18th International Conference on