Stochastic inversion for microwave imaging of dielectric bodies in the presence of noise

A new approach to microwave imaging of dielectric bodies is presented. The proposed procedure is based on the stochastic inverse of a linear system matrix derived from the coupled nonlinear integral equations used in microwave imaging by employing an equivalent current density. The prior knowledge required in this method is provided by a first estimate of the reconstructed values obtained by applying the Tikhonov regularization. It is found that, by the proposed method, the dielectric permittivities of two-dimensional bodies can be reconstructed with very high accuracy, in a reduced number of iterations, even for measured field data containing a high level of noise.