Comparative Study On EIT Static Reconstruction Algorithms Using The L1 Or L2 Norm

Electrical Impedance Tomography(EIT) is developed as a novel electrical measurement technique in recent years. The advantage of EIT is visualization, non-radiative, non-invasive, quick response, simple structure and low cost. EIT includes dynamic imaging and static imaging according to different imaging goal. Static imaging is aimed at reconstructing absolute conductivity distribution. Compared with dynamic imaging, imaging result contains much more information and has important implications for theoretical research and clinical diagnosis. Therefore static reconstruction algorithms become the focus of EIT technique. The objective function of imaging algorithm contains two terms, one is function of residual error between measurement boundary voltage and computational voltage, the other is penalty function. These two terms can be L1 norm or L2 norm respectively. In this paper, imaging ability and anti-noise property of algorithm using L1 norm or L2 norm is studied. Two terms both use L2 norm in Newton-Raphson algorithms. Algorithms which either one term or both terms use L1 norm can be solved using primal-dual interior-point method. The main work is as follows.1. The evaluation system is designed using simulation software Eidors-v3.7.1. It includes nine evaluation models and evaluation parameters. The evaluation parameters contain image correlation coefficient, image error and run time. The reconstruction ability of object’s position, shape, size, number and conductivity can be evaluated using this evaluation system.2. The principle and characteristic four Newton-Raphson static imaging algorithms is described and summarized. Images are reconstructed using these four algorithms. The evaluation parameters are calculated. Through the analysis of result, reconstruction ability of different algorithm is evaluated and compared.3. Primal-dual interior-point method is introduced and used to solve algorithms which either one term or both terms use L1 norm. Then images are reconstructed and the evaluation parameters are calculated. Through the analysis of result, reconstruction ability of different algorithm is evaluated and compared. Further, imaging ability and anti-noise property of L1 norm and L2 norm is studied.