Multi-frequency identification of defects in conducting media

An eddy currents based procedure for the 3D image reconstruction of defects in metallic plates from multi-frequency data is presented. In particular, we exploit the collection of data at different probe positions as well as at different excitation frequencies in order to improve the amount of information content, the accuracy of the inverse methodology and its robustness against the experimental noise. The identification tool we developed, exploits the geometric A − χ formulation for the solution of the eddy-current forward problem together with a full nonlinear iterative inversion algorithm based on the total variation regularization.