Assessment of Capability of the Target Decomposition Features in Sea Oil Pollution Detection Using Polarimetric SAR data

In this paper, the potential of polarimetric SAR data are evaluated using target decomposition features in order to detect oil spill are evaluated. For this purpose, the value of Power of all of the polarization channels (HH, VV, HV and VH) and decomposition features i.e. Freeman, Yamaguchi, VanZyl and Krogager are computed. Then, to detect oil pollutions, different combinations of these features are then classified using the support vector machine classifier. Thus, in addition to evaluating the ability of each of the used target decomposition methods, the most appropriate method for oil pollution detection is identified. The proposed method is conducted on full polarimetric SAR data acquired from PALSAR sensor of ALOS satellite. Evaluation results indicated that VV polarization is more proper compared to HH polarization for oil spill detection. Among the target decomposition methods, Krogager method with the overall accuracy of 97.3% provides the best result in this field.