Oil spill detection using marine SAR images

A probabilistic approach to distinguish oil spills from other similar oceanic features in marine Synthetic Aperture Radar (SAR) images has been developed and tested. The method uses statistical information obtained from previous mesurements of physical and geometrical characteristics for both oil spill and natural features. A sample image is evaluated using two different procedures to determine the probability that it is an oil spill, the results of the two procedures are then compared. The classification-algorithm performance was evaluated using a test dataset containing 80 examples that were oil spills and 43 that were natural features exhibiting characteristics similar to oil spills: more than 80% of the samples were classified correctly. The reliability of the method was then determined using a new dataset and similar results were obtained.