Floating target abnormal detection based on three united features

It is a challenge for marine surveillance radar to detect floating targets in sea clutter. In this paper, attentions have been focused on abnormal detection of floating targets on sea surface. We make a study of the nature of both sea clutter and target backscatter, and extract three basical features: Doppler centre frequency, the second order central moment and average power. Based on those three features, their distributed characters and diffusion properties are analyzed. It is found that target can not be distinguished from clutter with only one feature or two. Due to the failures of a one-feature-based detector and a two-feature-based detector, naturally, a three-feature-based detector is proposed as an abnormal algorithm. It is simple and heuristic, and is validated by real-life IPIX radar datasets. Experimental results show that the three-feature-based detector outperforms the compared detectors for the floating target on sea surface.