Application of GVSPM to Reconstruction of Solid-Air Two-Phase Flow CT Images (Comparison with Landweber and Iterative Tikhonov Regularization Methods)

Abstract

A new reconstruction method called Generalized Vector Sampled Pattern Matching (GVSPM) has been applied to an ill-posed inverse problem involving an electrical capacitance CT for solid air two-phase flow. This new mehod is able to achieve stable convergence without the use of an empirical value. This accurate reconstruction is accomplished using an objective function that is calculated as the inner product calculation between the experimental capacitance and the reconstructed image capacitance. The GVSPM method is compared with the conventional Landweber (LW) and Iterative Tikhonov regularization (ITR) methods in terms of capacitance residual, image error and image correlation. Overall, the accuracy is strongly dependent upon the image type and the iteration number, however the GVSPM method proved superior to the LW and the ITR methods in the case of annular pseudo particle images.