SAR image simulation of ocean environment and detection of oil slicks

Abstract:
In this paper Synthetic Aperture Radar (SAR) images of open sea including oil slicks are simulated and an algorithm to detect oil spill using morphological operators and a dynamic threshold is proposed. The simulation is based on a fractal model of the sea surface and includes the possibility to set wind direction and intensity, polarization, carrier frequency, incidence angle, spatial resolution and the approximate shape and size of the oil slick. The detection algorithm includes a block-wise detection that updates the backscattering threshold through a Gaussian Mixture Model used to separate oil from sea probability density function and a filtering based on morphological operator to refine the search of oil cells on the image. The precision and accuracy of detection are provided on images simulated in various operating conditions. Visual inspection has been used to evaluate the results on a real dataset taken from TerraSAR-X.