Modeling High-resolution SAR Images with Generalized Extreme Value Distribution

It is very important to develop precise statistical model for the high-resolution SAR images. The accurate distribution modeling of SAR images is the main contribution of this paper. In this paper, we study the GEV distribution model which can be used for modeling high-resolution SAR images, and introduce the properties and parameter estimation methods. In order to quantitatively assess the fitting result, we adopt the Kullback-Leibler divergence and the mean squared error as a similarity measurement. Experiment results with high-resolution SAR images indicate that the proposed GEV model can achieve satisfactory performance improvement in describing the statistical distribution of SAR image, not only in the homogeneous area, but also in the heterogeneous area.