Using UV Technique and DNN for the Profile Reconstruction of the Pierson-Moskowitz Rough Surface

In this paper, an inverse scattering scheme for the profile reconstruction of the Pierson-Moskowitz rough surface is proposed. The method of moments (MoM) based UV technique is used to generate the scattered fields efficiently. And the inverse model uses a deep neural network (DNN), whose input data are scattered fields, is implemented to reconstruct the rough surface profile. Numerical results show that the proposed scheme is capable of reconstructing the rough surface profile efficiently and accurately, with a relative error under 20%