A numerical assessment of the semiconvergence behavior in an inverse-scattering approach to electromagnetic imaging

In this paper an approach to invert measured data in electromagnetic imaging based on inverse scattering is numerically evaluated. In particular, an Inexact Newton approach is adopted and the semiconvergence behavior of the proposed iterative method, due to the ill-posedness of the considered inverse problem, is assessed: Indeed the iteratates of the algorithm first converge towards the exact solution up to a certain iteration number and, after that, the noise starts to dominate the restoration, and the subsequent iterations deviate from the exact one. Basic theoretical results concerning this semiconvergence phenomenon are briefly recalled, whereas a numerical assessment is performed concerning the application of the method to the image reconstruction of dielectric targets by using interrogating waves in a multiview arrangement.