|
Polarimetric synthetic aperture radar (PolSAR) serves as a crucial instrument in marine remote sensing. Making use of polarimetric information is paramount to enhancing the performance of PolSAR ship detection. The challenge lies in sufficiently utilizing comprehensive polarimetric information to simultaneously achieve superior false alarm suppression and ship detection. In this vein, this work dedicates to this issue and develops a novel PolSAR ship detection approach that employs multi-dimensional information fusion. The main contributions contain two aspects. Firstly, a novel polarization-space-time context covariance matrix (PSTCCM) within a local spatial neighborhood of sublook PolSAR images to characterize the target whole information is developed. This matrix amalgamates multi-dimensional information, including space, time, and polarization, derived from PolSAR data. Secondly, a similar pixel number (SPN) indicator based on PSTCCM that can effectively identify salient targets is further derived for ship detection. The underlying principle is that ships and sea clutter candidates exhibit different properties of homogeneity within a moving window, and the SPN indicator can clearly reflect these differences. The sensitivity and efficiency of the SPN indicator is examined and demonstrated. Comprehensive comparison studies are conducted using GaoFen-3 and Radarsat-2 PolSAR datasets. Quantitative investigations in terms of the figure of merit (FOM) index validate the superiority of the proposed method, especially for inshore false signals discrimination.
|
