Contrast of Contextual Fisher Vectors for Ship Detection in SAR Images

The problem of ship detection based on superpixels in synthetic aperture radar (SAR) images is considered. Existing superpixel-based detectors are often developed according to the simple contrast between the gray-levels of ship targets and the sea clutter surrounding it. These detectors are only based on first-order statistics of superpixels and may suffer from deteriorative performance in low signal-to-clutter ratio (SCR) scenarios because of the blurry SAR images. In this paper, we propose a new detector for ship targets in SAR images based on the contrast of contextual Fisher vectors (CCFVs). CCFV provides more abundant contrast between ship targets and their surrounding background clutter on multi-order information and multi-Gaussian components. Experimental results on the measured SAR images show that, compared with commonly used detectors, the proposed CCFV-based detector achieves better performance especially for weak target detection in low SCR scenarios.