Computational technique based on a Radial Basis Function network for microwave imaging of two- dimensional dielectric scatterers

In this paper, an innovative computational approach based on the use of a radial basis functions (RBF) neural network is presented for the solution of the inverse-scattering problem arising in microwave-imaging applications. In such a framework, this paper is aimed at assessing the effectiveness of the proposed approach in, shaping, and reconstructing the dielectric parameters of unknown scatterers from the knowledge of the scattered electric field in a two-dimensional geometry. The scattering electric fields of the training sets that characterize the objects are first computed by the method of moments (MoM). These sets are then utilized to effectively train a radial basis function (RBF) neural networks to accurately reconstruct the shape and determinate the permittivity profile of any unknown scatterer. The potential of the proposed approach is demonstrated in the case of a reconstruction of dielectric cylindrical objects. The neural results are in very good agreement with the theoretical and experimental results.