Signal and image processing for electrical resistance tomography

Electrical Resistance Tomography (ERT) is in essence an imaging technique.In ERT current is injected into and removed from a vessel via paired electrodes. The resulting voltage measurements are captured between the remaining electrode pairs. The principle behind ERT is to map these boundary voltages into a conductivity distribution that represents the domain of the vessel. The author has coded a versatile reconstruction algorithm based on the Newton-Raphson algorithm. The knowledge gained by implementing the algorithm is documented in this thesis. The literature covers the basic aspects of two-dimensional and three-dimensional ERT. It is hoped that this thesis will create a greater interest in ERT at the University of Cape Town (UCT) and also act as a building block for further developments. The thesis starts by presenting the basic concepts of ERT such as the underlying equations, the various boundary measurement strategies and a global perspective of ERT. The nature of this thesis is on software reconstruction and in so doing information on the incorporation of the Finite Element Method in ERT is provided. The thesis goes on to provide information about the reconstruction algorithms, which incorporate regularization. A novel aspect of this thesis involves the calibration and pre-processing of boundary voltages. These concepts were conceptualised and developed during formal communications with Dr. Wilkinson (2002) and Randal (2002). The calibration schemes try to eliminate the potential errors that can arise inthe captured data thus allowing for a clearer image to be reconstructed, Electrical Resistance Tomography. This thesis further develops the idea of parallelizing the Newton-Raphson algorithm to increase the speed of the algorithm. Various schemes on how this parallelization is achievable are put forward.

Grantor dc:publisher.institution

Department of Electrical Engineering

Year dc:date.issued

2002

Author dc:creator

• Naidoo, Thoneshan

Advisor dc:contributor.advisor

• Wilkinson, Andrew John

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/5140

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/5140