Target detection in non-stationary clutter background and Riemannian geometry

This study combines a smooth prior long autoregressive (SLAR) model and Riemannian geometry method to realise target detection in the presence of non-stationary clutter. First, SLAR is used for parameterisation of the signal. Then, the signal is mapped to a parameter vector space which can be described as a complex Riemannian manifold. Each point of this manifold is identified by a vector of AR coefficients. The principle of detection is that if a location has an enough Riemannian distance from the Riemannian median estimated by its neighbouring locations, targets are supposed to appear at this location. Numeric experiments and real radar target detection within sea clutter are given to demonstrate the effectiveness of the proposed targets detection method.