Microwave imaging from experimental data within a Bayesian framework with realistic random noise

This paper deals with the reconstruction of three-dimensional targets from experimental multiple-frequency data measured in the anechoic chamber of the Institut Fresnel (Marseille, France). An inverse iterative scheme is implemented with an adequate choice of the cost functional. Indeed, a Bayesian approach is considered in order to take into account the random noise which is present in the experiment. This leads to the definition of an adequate cost functional, where the weighting coefficients are changing with the frequency, the incidence angle and the receiving angle. The inversion scheme turns out to be more robust and accurate.