SAR Image Segmentation and Denoising Simultaneously Using Level Set Methods

In this paper, SAR image enhancement and segmentation have simultaneously been considered. Speckle noise inherent in SAR images, is the most important problem for segmentation of these images. Speckle is a multiplying noise with high damaging power which produces intensity inhomogeneity. As a result, usual segmentation methods cannot be applied on SAR images. In this paper, a combinational method is used for SAR image segmentation in order to benefit from the advantages of the pervious techniques and improve their performance. The method uses a suitable feature based on normalized cumulants which is independent of pixel intensity. Also, an image base method is used to model the intensity inhomogeneity resulting in speckle noise reduction. To evaluate the proposed algorithm, it has been tested on a set of simulated and real SAR images. Experimental results applied both on synthetic and real images demonstrate the effectiveness of proposed method.