Target Detection Based on Sea Clutter Model Using Neural Network

Abstract

A novel method to detect small target embedded in sea clutter is presented for high frequency (HF) radar. The method is rooted in different characters between instantaneous radial velocity of sea current and moving target, and relies on the neural network for its implementation. By estimating the instantaneous velocity of sea current and target, we find that a spatial nonlinear model rather than deterministic chaos model is more appropriate to describe the relationship among radial velocities of neighbor sea areas. Then we built a neural network model to approach to a predictor for sea clutter. So an incoming target will be detected for its more predicted error than a certain threshold. The method performs well on ocean echo data acquired by the HF radar system OSMAR2003.