Usability of Medium Resolution Optical Remote Sensing Images for Anomaly Detection in Maritime Surveillance Applications

Abstract

As part of the project "Intelligent Assistance and Analysis Systems for Early Detection and Management of Maritime Hazardous Situations” (IntelliMar) an anomaly detection application was developed and validated based on the analysis of Automatic Identification System (AIS) and Earth Observation (EO) remote sensing data. For this task optical Earth observation medium resolution satellite data from Landsat-8 and Sentinel-2 were used and their suitability in the context of object detection was evaluated. In a two-step approach, deep-learning methods were used for object detection and classification, and the derived results were then applied to a set of anomaly rules for anomaly report generation and transmission.