Register or Log In

International Society for Industrial Process Tomography

8th World Congress on Industrial Process Tomography

A Simulation Based Benchmark Dataset for Electrical Capacitance Tomography Image Reconstruction

Jin Zheng 1, Yi Li 2, Lihui Peng 1

1 Tsinghua National Laboratory for Information Science and Technology, Department of Automation, Tsinghua University

Tsinghua University, Haidian District, Beijing 10084, P. R. China summervega@yeah.net; lihuipeng@tsinghua.edu.cn

2 Graduate School at Shenzhen, Tsinghua University Tsinghua Campus, The University Town, Shenzhen 518055, P. R. China

liyi@sz.tsinghua.edu.cn

ABSTRACT

Electrical capacitance tomography (ECT) is a measurement technique for visualizing dielectric multi- phase flow processes, such as pneumatic conveying systems and fluidized beds, by generating cross- sectional images. ECT has developed decade years and made remarkable achievement, however there is still no public benchmark dataset for evaluating and comparing different image reconstruction algorithms for ECT. This paper builds a simulation based benchmark dataset for ECT image reconstruction, which contains 80 capacitance vectors including the condition of empty and full pipe for calibration, two sensitivity matrices for 8-electrode sensor and 12-electrode sensor respectively, and 36 phantoms of permittivity distribution of four typical two-phase flow patterns with different relative permittivity values and phase concentrations for each pattern. This benchmark dataset can be used to test image reconstruction algorithms, help researchers concentrate more efforts on the study of image reconstruction algorithm. It is helpful for the society to investigate different image reconstruction algorithms thus the reconstructed results can be evaluated and compared, which makes ECT image reconstruction research area more open and flexible, and hopefully brings breakthrough in ECT field.

Keywords benchmark dataset, electrical capacitance tomography, evaluation criteria, image reconstruction

Sign-in to access the full text