Advanced Oil Spill Detection Algorithms for Satellite based Maritime Environment Monitoring
Radius, Andrea; Azevedo, Rui; Sapage, Tania; Carmo, Paulo

During the last years, the increasing pollution occurrence and the alarming deterioration of the environmental health conditions of the sea, lead to the need of global monitoring capabilities, namely for marine environment management in terms of oil spill detection and indication of the suspected polluter.

The sensitivity of Synthetic Aperture Radar (SAR) to the different phenomena on the sea, especially for oil spill and vessel detection, together with its independence from the existing meteorological conditions and from the existence of light, makes it a key instrument for global pollution monitoring.

The aforementioned advantages of SAR in maritime pollution monitoring are being operationally explored by a set of service providers on behalf of the European Maritime Safety Agency (EMSA), which has launched in 2007 the CleanSeaNet (CSN) project - a pan-European satellite based oil monitoring service.

EDISOFT, which is from the beginning a service provider for CSN, is continuously investing in R&D activities that will ultimately lead to better algorithms and better performance on oil spill detection from SAR imagery. This strategy is being pursued through EDISOFT participation in the FP7 EC Sea-U project and in the Automatic Oil Spill Detection (AOSD) ESA project.

The Sea-U project has the aim to improve the current state of oil spill detection algorithms, through the informative content maximization obtained with data fusion, through the exploitation of different type of data and sensors and through the development of advanced image processing, segmentation and classification techniques, with the aim to implement more accurate algorithms for oil spill detection

The AOSD project is closely related to the operational segment, because it is focused on the automation of the oil spill detection processing chain, integrating auxiliary data, like wind information, together with image and geometry analysis techniques.

The synergy between these different objectives (R&D versus operational), allowed Edisoft to develop oil spill detection software, that combines the operational automatic aspect, obtained through dedicated integration of the processing chain in the existing open source NEST software, with the advanced algorithm development based on Wavelet filtering, for the improvement of oil spill detection and classification.
In this work we present the functionalities of the developed software and the main results in support of the developed algorithm validity.