Near Real Time Processing Framework for Remote Sensing Based Maritime Surveillance Applications

摘要

Applications for near real time object detection, based on remote sensing satellite image data, were developed and implemented as one of the first value adding services for Maritime Situation Awareness. The service was successfully demonstrated by different projects and is currently implemented for the SAR satellite sensors TerraSAR-X, TanDEM-X, Radarsat-2 and the Copernicus mission Sentinel-1; high resolution (HR) optical satellite sensors Landsat-8 and Copernicus Sentinel-2; and very high resolution (VHR) optical satellite sensors GeoEye-1, WorldView-1-4 and Deimos-2. Using TerraSAR-X and TanDEM-X satellites, the service is available for the commercial market via AIRBUS DS. Service based on very high resolution optical data is developed in partnership with European Space Imaging (EUSI), key provider of VHR optical satellite imagery in Europe. With respect to user needs different product formats are supported and operationally available. Fusion with AIS data from terrestrial and satellite AIS sources is integrated to provide a more reliable picture with information like ship name, ship class, and destination. The SAR algorithms are developed by the Maritime Security Lab in Bremen, part of the Remote Sensing Technology Institute (IMF), while the Maritime Security Lab in Neustrelitz is responsible for optical data processing as well as the near real time framework development and the service chain implementation at DLR's Ground Stations. The Maritime Safety and Security Labs in Bremen and Neustrelitz were established at DLR's Earth Observation Center (EOC), focusing on the research and development of new EO near real time services in the maritime domain. Beside ship detection services the development of oil spill detection, iceberg detection and ice classification, as well as the delivery of products of meteo-marine parameters like wind and wave are currently part of the research and development activities. This paper will focus on the frame work implementation as an