A Novel Level Set Method for Inhomogeneous SAR Image Segmentation

Abstract

Segmentation is an essential part in SAR image analysis and application. However, the presence of speckle noise and intensity inhomogeneity regions leads to great difficulty in SAR image segmentation. In this letter, we propose a novel level set method that integrates the local intensity and the global feature information for better segmentation results. To establish the energy function, a new image model is developed to describe the inhomogeneous SAR images and form the local intensity clustering term. Meanwhile, the feature map extracted by invariant moment is used to define the global fitting term through a piecewise constant function. Segmentation is realized by solving the level set function from the minimization of the energy function. Experiments on the synthetic and real SAR images verify the effectiveness of the proposed method.