Comparison of Three Image Reconstruction Algorithms for Electrical Capacitance Tomography

S Liu1, L Fu2 and W Q Yang3

1 Institute of Engineering Thermophysics, Chinese Academy of Sciences, PO Box 2706, Beijing 100080, PR China, Email: shiliu@hotmail.com

2 Department of System Engineering and Applied Mathematics, Chongqing University, Chongqing 400044, PR China, Email: fuli@dns1.cqu.edu.cn

3 Department of Electrical Engineering and Electronics, Process Tomography Group,

UMIST, PO Box 88, Manchester M60 1QD, UK, Email: w.yang@umist.ac.uk

ABSTRACT

This paper compares linear back-projection (LBP), iterative and generalized inverse or Moore Penrose Inverse (MPI) methods for electrical capacitance tomography (ECT), together with experimental evaluation. It is concluded that (1) LBP is incapable of distinguishing multiple-objects in close vicinities, (2) while MPI analytically produces the least error, it is not suitable for image reconstruction, and (3) the iterative method is powerful in reconstructing complex object distributions. However, uncontrolled iterations trend to server distortions and the reconstructed images would approach those reconstructed by the MPI method.

Keywords Linear back-projection, Iterative, Generalized inverse, Image reconstruction