A Regularization Structure Based on Novel Iterative Penalty Term for Electrical Impedance Tomography

This study proposes an iterative penalty term-based regularization (IPTR) for electrical impedance tomography (EIT) imaging. A penalty term in regularization is formed based on the supergradient of a concave function to eliminate the smooth-edge effect of reconstructed images. Compared with other variant regularization methods, the calculation of IPTR method is simpler and does not introduce additional undetermined parameters. To evaluate the effectiveness of the IPTR method, numerical simulations and tank experiments are carried out. Results indicate that compared to typical Tikhonov regularization and L1-norm regularization, IPTR has greater stability and imaging precision (up to 26% accuracy improvement). Additionally, IPTR is capable of rapidly iterating (no more than 4 times) while maintaining a balance between image fidelity and sparsity, which can satisfy efficiency requirements in the application.