Oil spill detection in SAR images using minimum cross-entropy thresholding

Oil spill detection becomes very important nowadays, due to the importance of oil and to prevent pollution caused by oil leakage. Synthetic Aperture Radar (SAR) images show oil spill boundaries clearly. An oil spill appears as an obscure spot in SAR images. Therefore, thresholding is considered the best method to extract this patch and define its boundaries clearly. This paper proposes a novel thresholding method for detecting oil spill in SAR images that minimizes cross-entropy between images and their segmented versions using gamma Distribution. Moreover, this paper demonstrates applying the new proposed method results on artificial images as well as various SAR images. Gamma Distribution was chosen over other distributions because it has a general shape, and it showed excellent results in modeling images´ data.