Ship Target Detection Method in SAR Imagery Based on Generalized Pareto Manifold

Abstract

Ship target detection based on Synthetic aperture radar (SAR) imagery is a challenging task. With the continuous improvement of SAR resolution, the false alarm rate set by manual experience in traditional CFAR detection tends to lead to missed detection and weaken the detection performance. To solve this problem, this paper proposes a ship target detection method based on the generalized Pareto manifold in SAR imagery. The generalized Pareto distribution family is used to construct the SAR images’ statistical manifold, and the tangent vector length is applied to represent the local neighborhood of each pixel of SAR images, which indicates the difference between targets and the background clutter significantly, implementing the precise positioning of ship target and the effective suppression of background clutter in SAR images. The test results of Gaofen-3 satellite data show that compared with the traditional CFAR algorithm, this method achieves significantly better performance.