Scattering center estimation of HRRP via atomic norm minimization

As the one-dimensional projection of the echoes from scattering centers on a target along the radar line-of-sight, high range resolution profile (HRRP) has drawn great attention to automatic target recognition (ATR). Normally, features are extracted based on the strong scattering centers of HRRP. In recent years, several scattering center estimation methods have been proposed, such as Prony and orthogonal matching pursuit (OMP). Prony needs the number of scattering centers a priori and is sensitive to noise. Due to discretization, OMP may lead to the off-grid effect. In this paper, we propose to estimate the scattering centers via atomic norm minimization (ANM). Similar to OMP, ANM is a sparsity based approach, but the signal is modelled in continuous space, thus alleviating the basis mismatch. Simulation and experimental results demonstrate that ANM achieves higher accuracy in terms of both position and amplitude compared with OMP.