Ship Detection in Polarimetric Sar Image Based on Similarity Test

Ship detection is an important application in polarimetric synthetic aperture radar (PolSAR) image. A novel saliency detector for PolSAR ship detection has been proposed in this work, which considers ship targets as salient candidates from sea clutter in low and medium sea conditions. Firstly, the similarity test based on polarimetric covariance matrix is applied on each pixel within its neighborhood. Then, saliency feature named Similar Pixel Number (SPN) is generated by calculating the similar samples within its neighborhood. Finally, ship targets can be obtained through proper thresholding procedure and morphological filtering. Experimental studies with two Radarsat-2 datasets validate the advantages of the proposed method.