Small target detection in SAR image using the Alpha-stable distribution model

Abstract

The Constant False Alarm Rate (CFAR) algorithm is most commonly used for small target detection in SAR images. As the goodness-of-fit of distribution model to SAR clutter has great effect on the performance of algorithm, after a comprehensive statistical analysis of background clutters of different SAR data, a modified CFAR algorithm based on the Alpha-stable distribution is proposed for detecting small targets in SAR images, especially under the extremely inhomogeneous background clutter. Considering for the complexity of Alpha-stable distribution model, the parameter estimation and threshold determining steps of the modified algorithm are introduced in detail. Performance of the algorithm is assessed by experiments on ADTS data. Compared with typical two-parameter CFAR (TP-CFAR) algorithm based on Gaussian distribution and K-CFAR algorithm based on K distribution, the proposed method is demonstrated to be most suitable for detecting small target in extremely inhomogeneous regions.