A Sparsity Reconstruction Algorithm of Electromagnetic Tomography Technique for High Conductivity Medium Imaging

Abstract

Abstract Electromagnetic tomography (EMT) is a versatile tomographic imaging technique for reconstruction of conductivity and/or permeability distribution due to the advantages of non-contact, non-intrusive, low-cost, simple structure and fast imaging. However, the ill-posed and ill-conditioned features of EMT make it difficult to obtain high quality reconstructed images. To improve the spatial resolution of the high conductivity medium imaging, the L1-L1 framework objective function is presented, which uses L1 norm as both data fidelity term and regularization term to weaken the influence of the data outliers and impose the sparsity feature of the measured objects. An improved Split Bregman method is proposed to solve the complicated optimization problem efficiently, which splits it into several simple sub-tasks. Each subtask can be solved by adopting the proper method. Besides, an acceleration strategy is introduced to improve the convergence rate. Numerical simulations are used to verify the effectiveness and competitive performance of the proposed improved method. The experiments are carried out by the designed modularized EMT system to further verify the effectiveness of the proposed method. The reconstructed images can precisely show the number and positions of the measured objects.