Adaptive $l_{p}$ Regularization for Electrical Impedance Tomography
Jia Li,Shihong Yue,Mingliang Ding,Ziqiang Cui,Huaxiang Wang
DOI: https://doi.org/10.1109/jsen.2019.2940070
IF: 4.3
2019-01-01
IEEE Sensors Journal
Abstract:Owing to its low cost, fast response, non-invasiveness, and non-radiation, electrical impedance tomography (EIT) has been applied to numerous fields. However, its spatial resolution is low due to the inherent ill-posed problem and the "soft field" effect. The L_{p} regularization ( 0 < \textit {p} < 2 ) is effective for overcoming these disadvantages, and efforts have been made to use regularization from the most popular L_{2} to its variants L_{1} and L_{1/2} . Nevertheless, L_{p} regularization is generally difficult to be solved fast and efficiently, and the selection of p yielding the best result is also a problem. In this paper, an adaptive re-weighted (ARW) algorithm with a general frame is presented to solve the L_{p} regularization for EIT, with p for each pixel determined adaptively in iterations. Experiments were carried out to validate the proposed algorithm. Results show that compared with other EIT algorithms, the ARW algorithm had a higher spatial resolution. Moreover, it can provide a wider range of selection for regularization parameter, which increases the practicality of the proposed algorithm.