Weigthing hyperparameters for 3D Bayesian estimation in Eddy current tomography

The skin effect is a major limitation for Eddy Current Tomography ECT. It limits its ability to image flaws inside electricaly conductive material. In this paper we tackle the inverse problem in a Bayesian framework which imply the computation of the hyperparameters defining the weight of the a priori in the estimate. To improve its performance regarding buried flaws, we suggest that hyperparameters of the inverse problem can't be the same for each layer of discretization and we propose a method to bias them accordingly to the skin effect.