Abstract: Polarimetric synthetic aperture radar (SAR) has been widely used in maritime ship detection due to its abundant target scattering information. In this paper, the visual saliency region extraction method based on residual spectrum in machine vision is extended to Pol-SAR ship detection. Firstly, the ship discrimination ability of SAR polarimetric features is analyzed, and 5 RADARSAT-2 full-polarization data are used to compare with the Euclidean distance, Bhattacharyya distance and ship-sea contrast of 45 polarization features, then, 16 polarimetric features with the contrast greater than 20 dB are selected. Secondly, the combination of polarimetric features suitable for the saliency regional method of residual spectral is picked out. Based on the analysis of the mutual information of features and the logarithmic spectrum characteristics of the image, three amplitude features of the coherent matrix T are selected and combined into RGB images for ship detection. Finally, the method presented is compared with the traditional constant false-alarm rate (CFAR) method. The figure of merit of the method in this paper is 0.95 while the value of traditional CFAR method is less than 0.9. The result shows that the false alarm can be restrained well, and geometric information such as the contour of target can be directly extracted, the method has certain application prospect.