Premium
Regularized modified Newton‐Raphson technique applied to electrical impedance tomography
Author(s) -
Grootveld C. J.,
Segal A.,
Scarlett B.
Publication year - 1998
Publication title -
international journal of imaging systems and technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.359
H-Index - 47
eISSN - 1098-1098
pISSN - 0899-9457
DOI - 10.1002/(sici)1098-1098(1998)9:1<60::aid-ima8>3.0.co;2-p
Subject(s) - electrical impedance tomography , computer science , pixel , property (philosophy) , finite element method , newton's method , algorithm , process (computing) , electrical impedance , eigenvalues and eigenvectors , mathematical optimization , computer vision , mathematics , electrical engineering , engineering , nonlinear system , physics , quantum mechanics , philosophy , structural engineering , epistemology , operating system
Electrical Impedance Tomography (EIT) is gaining importance as a monitoring tool for process engineering. The main reasons for this are its nonintrusive measurement property and its relatively cheap hardware. However, the image reconstruction still remains a problem especially under heavy process conditions with little prior information. Many researchers have devoted their attention to this problem, but robust algorithms working on real noisy data are scarce. The authors present a regularized, modified Newton‐Raphson algorithm that gives satisfying results on both static and dynamic processes. The number of pixels used by the algorithm are identical to the number of true non‐zero eigenvalues, thereby diminishing the effect of ill‐conditioning. The algorithm uses a user‐defined pixel mesh that is mapped onto a standard finite‐element mesh so that the user is able to easily adapt the mesh to the problem under investigation. © 1998 John Wiley & Sons, Inc. Int J Imaging Syst Technol, 9, 60–65, 1998