Published in IET Science, Measurement and Technology
Received on 31st July 2013
Revised on 18th December 2013
Accepted on 7th March 2014
doi: 10.1049/iet-smt.2013.0124
ISSN 1751-8822
Three-dimensional electrical capacitance tomography
reconstruction by the Landweber iterative algorithm
with fuzzy thresholding
Hua Yan, Yi Fan Wang, Ying Gang Zhou
School of Information Science and Engineering, Shenyang University of Technology, Shenyang 110870,
People’s Republic of China
E-mail: yanhua_01@163.com
Abstract: The image reconstruction for electrical capacitance tomography is a non-linear, underdetermined and ill-posed inverse
problem. It is difficult to obtain a reconstructed image with high quality, especially in the case of three-dimensional (3D)
reconstruction. An iterative image reconstruction algorithm with fuzzy thresholding is proposed in this study. The threshold
value in each iteration is generated by minimising the measure of fuzziness of current reconstructed image. The algorithm
proposed is tested by the noise-free and the noise-contaminated capacitance data. Extensive computer simulations demonstrate
that the fuzzy thresholding can reduce low grey-level artefacts effectively. As a result, the spatial error, volume error,
permittivity error and scattered artefacts of the reconstructed image are reduced obviously; not only that, the number of
iterations needed to obtain a good reconstruction result is decreased greatly. The result of 3D reconstruction of a H-shape
object verifies the effectiveness of the fuzzy thresholding further.
1 Introduction
Owing to the distinct advantages such as easy
implementation, low cost, high safety and non-intrusive
sensing, electrical capacitance tomography (ECT) is
considered as a promising visualisation measurement
technique, in which reconstructing high quality images is
highly desirable for real applications [1, 2]. ECT is based
on measuring capacitances of a multi-electrode sensor
surrounding a vessel or pipe containing materials of
different permittivities. These measurements are used to
reconstruct the two-dimensional (2D) or the 3D permittivity
distribution and hence the material distribution inside the
vessel by using a suitable algorithm. Numerous studies
have been realised in order to optimise the performance of
ECT system with the aim of improving the low resolution
that is inherent to the ECT technique [3, 4].
In the past few years, 3D ECT imaging [5–10] has gained
increased attention. A 3D ECT sensor usually has at least
two-plane measuring electrodes in axial direction. For a
fixed number of electrodes, the increase of planes is
countered by a decrease in radial resolution. In some
publications, 3D ECT is also called electrical capacitance
volume tomography [8, 9, 11] due to the fact that it can
realise volumetric imaging.
Successful applications of ECT depend greatly on the
precision and speed of the image reconstruction algorithm.
There are several difficulties associated with image
reconstruction in ECT [12–14]. (i) The relationship between
the permittivity distribution and capacitance is non-linear.
(ii) The sensitivity distribution of ECT sensor is
non-uniform and would change with different permittivity
distributions. (iii) It is an underdetermined problem since
the number of independent capacitance measurements is
much smaller than the number of desired image pixels and
(iv) It is an ill-conditioned problem; the condition number
of the sensitivity matrix is very large, resulting in the
magnification both of measurement errors and of numerical
errors in the reconstructed image. In the case of 3D ECT,
these difficulties become more serious since the sensing
field is more uneven and the number of pixels increases
many times while the independent measurements are
unchanged.
In general, image reconstruction algorithms for ECT can be
categorised into two groups: non-iterative algorithms and
iterative algorithms. The linear back projection (LBP)
algorithm is the most commonly used non-iterative
algorithm. It is very fast and can be used for on-line
reconstruction. Certainly, it has limited resolution. Owing to
the non-linear relationship between the permittivity
distribution and the capacitance, it is almost impossible to
find an accurate solution by any non-iterative algorithm
with a simplified linear model. To improve the image
quality, the inverse problem has to be solved iteratively.
Iterative algorithms developed for ECT are based on
calculating capacitance values from the permittivity
distribution of the current image, and then producing a new
image using the discrepancy between the measured
capacitance and the calculated capacitance. Various iterative
reconstruction algorithms have been developed, such as the
www.ietdl.org
IET Sci. Meas. Technol., 2014, Vol. 8, Iss. 6, pp. 487–496
doi: 10.1049/iet-smt.2013.0124
487
&
The Institution of Engineering and Technology 2014