Optimum detection and segmentation of oil-slicks with polarimetric SAR data

A new polarimetric discriminator, derived by using the generalised likelihood approach, is proposed in this paper for the detection of slicks on the sea surface. A complete analytical expression of the detection performance is derived for the proposed detector and used to compare it to other conventional polarimetric detectors, showing its better performance. In particular, the improvement obtained by using the polarimetric images with respect to the best single channel image is demonstrated. Moreover it is shown that the ML discriminant outperforms conventional polarimetric detectors. The results achieved in the segmentation of the SIR-C/X-SAR image of the experimental set up in the German Bight confirm the results of the theoretical performance analysis.