Image reconstruction algorithms using overcomplete dictionaries for ECT sensor
L. de Moura, D. R. Pipa, A. N. Wrasse, M. J. da Silva
Graduate Program in Electrical and Computer Engineering, Federal University of Technology - Parana – CPGEI/UTFPR
Av. Sete de Setembro, 3165 – Curitiba, Brazil email@example.com
In industrial applications such as oil production and transportation, safe operation often requires information about the contents inside a pipeline. Two-phase flows are quite common in oil transportation, as the oil rarely fills the entirety of the pipeline cross-section. Phase distribution is an important parameter which characterizes two-phase flows. Many techniques were developed to estimate the phase distribution and with it other information like void fraction, phase velocity and flow rate. Among these techniques is the Electric Capacitance Tomography (ECT), a non-invasive technique that uses electrical field changes to discern materials with different electrical permittivities. In past years, image reconstruction techniques with L1 regularization and sparse representations have been extensively studied. The use of overcomplete dictionaries has also been subject of many studies in recent years. The aim of this work is to investigate the effectiveness of image reconstruction algorithms using an overcomplete dictionary and L1 regularization to promote sparseness. To assess the proposed technique, we use a capacitive sensor array with 8 electrodes that is capable of making parallel measures at the cost of having reduced combinations of electrodes when compared to common ECT equipment. We present results with simulated and real data acquired in previous studies with this sensor. Early results show an overall improvement in the reconstructed images and in void fraction estimation.
Keywords Electrical Capacitance Tomography, Image Reconstruction, Overcomplete Dictionary, two-phase flow
Copyright © International Society for Industrial Process Tomography, 2016. All rights reserved.