Ship detection based on Radarsat-2 full-polarimetric images

Feature extraction is very important when processing the polarimetric synthetic aperture radar (SAR) image, and the key of ship detection. In order to enhance the contrast of ships and clutter, we investigate a new detection parameter (DP) for ship detection based on feature vector with three parameters: span value, the smallest eigenvalue and the combination of entropy and anisotropy, which have the highest TCR (Target-Clutter Ratio) when compared with other parameters after analyzing the Radarsat-2 full-polarimetric SLC data in our experiment. The results show that DP is very effective to enhance the TCR of the image and discriminate all the ships from the background ocean correctly and completely.