Image Filtering With Associative Markov Networks for ECT With Distinctive Phase Origins

The images reconstructed by electrical capacitance tomography (ECT) for two-phase flows are usually blurry at the phase interface. To improve the image quality, image filtering with associative Markov networks (AMNs), which support efficient graph-cut inference for insulation segmentation, is presented. An ECT sensor with 12 electrodes is investigated and the capacitance between different electrode pairs is calculated for some typical permittivity distributions using a finite element method. The initial images are reconstructed by liner back-projection and Landweber iterative algorithm, respectively. The obtained images are then processed using AMNs. Simulation results show significant improvement in the quality of images.