Multidimensional joint domain localized matrix constant false alarm rate detector based on information geometry method with applications to high frequency surface wave radar

Abstract

This paper proposes a joint domain localized matrix constant false alarm rate detector based on the space-time adaptive processing and information geometry method to improve the detection performance for high-frequency surface wave radar. Unlike traditional target detectors that use one-dimensional or two-dimensional amplitude information, the proposed detector considers multidimensional information of the signal to easily distinguish the target from the noise or clutter. The multidimensional information is obtained by joint domain localized processing, a reduced dimension space-time adaptive algorithm. Then using the information geometry method, we map the detection problem to the geometric manifold and detect the targets by geometric measures. This paper proposes different detectors based on different geometric measures. The experiments and results show that the proposed detectors can enhance the detection performance compared with classical cell average constant false alarm rate detector.