FINDINGS OF THE DECLIMS PROJECT - DETECTION AND CLASSIFICATION OF MARINE TRAFFIC FROM SPACE

The 24-partner DECLIMS project aims to establish and enhance the state-of-the-art in operational ship detection from space. Benchmark tests of automatic SAR detection algorithms have shown that existing systems are mature enough to be used operationally, but at the same time still suffer from problems that can be remedied by implementing available knowledge. Existing systems score less well on automatic vessel classification in SAR images; more fundamental improvements seem to be needed there. The state of development of automatic ship detection and classification algorithms for optical images lags behind that of radar. The characteristics of the satellite-based information (minimum detectable vessel size, area coverage, repeat frequency, etc.) and the way in which it can complement other information on ship traffic dictate the level of acceptance and maturity in the various application areas, ranging (in decreasing order) from fisheries control, via defence, pollution control, and maritime security, to traffic control. These application areas pose different requirements on future sensors; those from fisheries control are discussed in some detail. The DECLIMS project aims to establish the state of the art in detection and classification of maritime traffic from space, and to be a focus for research in that field. The project, partially funded by the European Commission, brings together 24 actors and stakeholders in satellite ship detection and classification, from a variety of backgrounds: small and large industry, institutes and academia; civilian and defence; from within the EU and outside. The project partners are listed in Fig. 1. Imaging sensors are considered, both optical and radar. The project started May 2003 and is planned to last 3 years. This paper intends to give a brief summary of the main findings of the project so far. The discussion is divided into optical results, radar results, users, data fusion and needs for future sensor systems. The primary interest is in the operational status of the methods, and less in theoretical or exploratory possibilities: what can really be done today with space-based earth observation sensors for maritime surveillance?