Oil slicks detection using a polarimetric region classifier

A new region based classifier for polarimetric synthetic aperture radar data (PolSAR) was tested to evaluate its potential to discriminate different types of oil slicks at sea surface. This classifier uses a supervised approach to compare stochastic distances between complex Wishart distributions and hypothesis tests to associate confidence levels to the classification results. The preliminary results using the Battacharyya distance were promising, returning an overall accuracy of 90.61% at a significance level of 5%. Future works may compare the performance of different stochastic distances, together with the insertion of polarimetric features to improve the oil slicks classification.