Comparison of TV and Tikhonov regularization in MIT for low and high conductivity imaging
Fang Li, M Soleimani
Engineering Tomography Laboratory (ETL), Department of Electronic and Electrical Engineering.
University of Bath, Claverton Down, Bath, BA2 7AY, United Kingdom firstname.lastname@example.org
Magnetic induction tomography (MIT) is an imaging technique with wide range of potential applications. As an imaging technique, MIT is aiming at visualizing the conductivity distribution of the object under test, which can be achieved by solving the inverse problem and forward problem. Inverse problem will be our main focus. Several inversion algorithms have been presented since the ill-posed and ill-conditioned property of inverse problem in MIT. This paper presents for the first time split Bregman total variation (TV) regularization to solve the MIT inverse problem. And the evaluation of the use of Tikhonov regularization algorithm and Split Bregman iterative algorithm for the inverse problem of magnetic induction tomography is presented as well. Tikhonov regularization is solved by l2-norm and TV is solved using the Split Bregman formulation, which has been shown to be optimal for l1-norm regularization. The feasibility of these two inversion solvers is critically evaluated by experimental data using high frequency and low frequency MIT data. Compared to the Tikhonov regularization algorithm for the conventional l2-norm, the total variation regularization method for the l1-norm can significantly enhance the imaging quality, which can be proved from a number of experimental results, which will make the TV method a great candidate for MIT imaging.
Keywords Eddy current forward problem, magnetic induction tomography, Split Bregman, Total variation regularization.
Copyright © International Society for Industrial Process Tomography, 2016. All rights reserved.