Combination of Statistical Similarity Measure and Derivative Morphological Profile Approach for Oil Slick Detection in SAR Images

Synthetic Aperture Radar (SAR) is widely used to detect and monitor oil pollution on the sea surface. As it is sensitive to surface roughness, the presence of oil film on the sea surface decreases the backscattering of this target type resulting in a dark feature patches in SAR images. In this paper, a new approach for oil slicks detection is presented. It is mainly based on SAR image texture analysis using the combination of a statistical similarity measure and a derivative morphological profile. Oil slicks signature is extracted trough two steps procedure. First, SAR image inspection is performed in order to highlight the dark spots suspected to be oil slicks. The inspection is achieved through a similarity measure between a local probability density function (lpdf) of clean water and the lpdf of the area to be inspected. The local distribution is estimated in the neighbourhood of each pixel and compared to a reference one using the Kullback-Leibler KL distance between distributions. Second, and once spots highlighted, texture features extraction using the Derivative Morphological Profile is porformed in order to improve the detection results. algorithm has been applied to Envisat Advanced Synthetic Aperture Radar (ASAR) and European Remote Sensing (ERS) images and it yields an accurate segmentation results. Indeed, the features extraction improves the detection slicks probability Pd of ASAR, respectively ERS, images from 93.08 %to 97.37 %and from 96.32 to 99.57 % on one hand, and reduces the false alarms probability respectively from 6.92 to 2.63 %and from 3.68 to 0.59 % on the other hand.