電気インピーダンストモグラフィの高分解能化に向けたポストプロセス開発

抄録

　We have developed an advanced post-processing technique designed to enhance the resolution of electrical impedance tomography (EIT). The technique used logical correlations between images reconstructed by several solving methods of inverse problem. We obtained various potential spectra using both adjacent and opposite measurement modes and visualized them using both regularization and machine learning methods. For object size, machine learning was the most effective. For object position, logical AND processing using Gauss-Newton method from both adjacent and opposite modes was effective.