Modification of CFAR Algorithm for Oil Spill Detection from SAR Data

Abstract

It is very difficult to detect oil spills when the scattering intensity of background clutter is inhomogeneous in synthetic aperture radar (SAR) images. To improve the oil detection capability, we propose a modified constant false alarm rate (CFAR)-based method for the detection of oil spills in SAR images. This proposed method combines edge detection technique and CFAR detection theory to improve the accuracy of oil spills detection. First, we segment the image into the areas of interest (AOIs) by using ratio edge detection. Second, to get a more accurate detection result, an improved Weibull-CFAR detector is applied to these AOIs. Experimental results demonstrate that the modified CFAR algorithm can work more effectively than a global CFAR detector for oil spill detection, especially for the inhomogeneous intensity SAR images. This model can detect the target more effectively, and false alarms can be greatly diminished.