An accurate oil slick detection method for SAR images

Oil slick cause severe ecological disasters and enormous marine environment damages. Due to the various sizes and irregular shapes of oil slick, the existing constant false alarm rate (CFAR) algorithms had poor detection performance for these multifarious areas. Therefore, an accurate detection method based on improved CFAR model is proposed in this paper. Firstly, global detection is utilized to acquire the areas of interests(AOIs). Then, in order to improve the estimation accuracy of the background probability density distribution, the small and closely located AOIs are treated as an entirety and are divided into one group. In addition, the rest AOIs are gathered into the other group. Finally, different types of AOIs are detected by sliding-windows with distinct size. Simulations and processed measured data illustrate that the proposed method has outstanding detection performance without increasing the computational complexity.