Weakly Convex Discontinuity Adaptive Regularization for Microwave Imaging

Reconstruction of inhomogeneous dielectric objects from microwave scattering is a nonlinear and ill-posed inverse problem. In this communication, we develop a new class of weakly convex discontinuity adaptive (WCDA) models as a regularization for quantitative microwave tomography. We show that this class includes the Huber regularizer and we show how to combine these methods with electromagnetic solvers operating on the complex permittivity profile. 2D reconstructions of objects from the Institute Fresnel database and experimental data at a single frequency demonstrate the effectiveness of the proposed regularization even when employing far less transmitters and receivers than available in the database.