Abstract
This paper describes an algorithm for electrical impedance imaging that makes no prior assumptions about current flow paths. It involves the solution of Poisson's equation for inhomogeneous media with an explicit conductivity-updating scheme that is not subject to matrix ill-conditioning problems. Sparse matrices are employed throughout and so a large number of pixels may be accommodated. A three-dimensional laboratory experiment, with top-surface measurements only, is described. A submerged metallic object is imaged with error believed due to poor modelling of fields in the vicinity of the physical electrodes. The algorithm requires a large number of conductivity-updating iterations and so schemes must be considered for a substantial reduction in this number.