Capacitance tomography has been used to image several processes, such
as liquid/gas pipe flow, oil/water/gas gravity separation, pneumatic c
onveying, fluidized beds and flame combustion. The nature of the capac
itance sensors is such that reconstruction algorithms well developed f
or medical tomography are not applicable. The main problem is that the
relationship between the measured quantity (capacitance) and the para
meter of interest (distribution of the dielectric constant) is nonline
ar. Furthermore, it is impossible to establish an explicit expression
which relates the dielectric constant distribution to the measured cap
acitance. Also it should be pointed out that the number of measurement
s in capacitance tomography is small (typically less than 100) compare
d to medical tomography. For these reasons the first tested algorithm
in capacitance tomography was based on the crude back projection algor
ithm. This algorithm has over the years been enhanced for use with a c
apacitance tomograph. In addition other techniques, such as various it
erative methods, algorithms based on artificial neural networks and 'l
ook-up' tables have been developed and tested. This paper outlines the
working principles for the different techniques and presents the main
results.