Detector

Linguistics

Pixel

Philosophy

Object detection

Similarity (geometry)

Polarimetry

Scattering

Telecommunications

Computer science

Thresholding

Geology

Physics

Pattern recognition (psychology)

Optics

Image (mathematics)

Synthetic aperture radar

Feature (linguistics)

Artificial intelligence

Remote sensing

Clutter

Radar

Computer vision

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

Ship detection in polarimetric SAR (PolSAR) image plays an important role in marine monitoring. From the viewpoint of the attention mechanism, ship targets can be salient candidates from sea clutter. A novel saliency detector for PolSAR ship detection has been proposed in this paper. The core idea is to explore the different scattering mechanisms between ships and sea clutter in low and medium sea conditions. The scattering mechanism differences are measured by the similarity test of polarimetric covariance matrices. A new saliency feature named similar pixel number (SPN) is proposed by recording the similar pixels within a moving window and a saliency detection method is developed thereafter. The proposed ship detection method mainly contains three steps. Firstly, the similarity test is applied between the central pixel and its neighborhood. Then, SPN feature is generated based on the result of similarity test by counting the number of similar pixels. Finally, with thresholding and morphological filtering procedures, ship targets can be extracted from the saliency feature map. Experimental studies with Radarsat-2 and GaoFen-3 data validate the advantages of the proposed method. Quantitative comparisons illustrate that the proposed method achieves the best detection performance with detection quality factors higher than 97% for both datasets over inshore dense and offshore sparse areas.

A Saliency Detector for Polarimetric SAR Ship Detection Using Similarity Test