Detecting motion anomalies
Kevin R. Keane · 2017
An unsupervised methodology is presented for the detection of motion anomalies using spatial context and multivariate statistical tests. The method is applied to GPS data captured for a taxi fleet in Porto, Portugal; and, AIS data captured for ships operating in the Aegean Sea. The autonomous method successfully identifies atypical trajectories. Currently under-exploited object motion data streams may yield further information from application of the proposed approach.