PolSAR Ship Detection Based on the Polarimetric Covariance Difference Matrix

Abstract

Ship detection using Polarimetric SAR data has attracted a lot of attention in recent years. Due to the sampling of the Doppler spectrum at finite intervals of the pulse repetition frequency, the azimuth ambiguities often appear in PolSAR images, which make the ship detection in PolSAR images frequently generating false alarms, especially in the case of low backscattering sea environment. In order to handle the problem and improve the performance of ship detection in PolSAR images, this paper presents a new method, which is mainly based on concentrating the polarimetric difference between ship pixels and background pixels. We first calculate a polarimetric covariance difference matrix, denoted as polarimetric covariance difference matrix (PCDM), by accumulating the elemental difference between the polarimetric covariance matrix at each pixel and the counterparts in its 3 × 3 neighbors. The SPAN detector is then applied on PCDM to obtain a coarse detection result. Meanwhile, we decompose the PCDM matrix to calculate a new polarimetric signature, called pedestal ship height (PSH), and use it together with the coarse detection result to distinguish ships from ambiguities. Extensive experiments on three real PolSAR datasets are carried out to demonstrate the effectiveness of the proposed method in comparing with other algorithms. The experimental results show that the proposed method not only detects ships effectively, but also can remove the azimuth ambiguities and reduce the false alarms significantly.