| Recently, Electrical Tomography (ET) develops quickly in industrial and biomedical areas due to its advantages-safety, high real time and low cost. Wire Mesh Tomography (WMT), an invasive technique, was widely applied in oil and chemical industrial areas, because of the simple measurement principle. However, restricted by the mesh size, the image accuracy can not be improved dramatically, especially for oil-water two-phase flow with high water hold up. So improving image resolution is becoming more and more important. Also, Electrical Impedance Tomography (EIT) is superior to CT and MRI techniques in clinic monitoring and early diagnose of tumor. Because the inverse problem is ill-posed, the stability and image accuracy are key research problems in EIT. Therefore,this thesis do a further deep research on image reconstruction algorithm. The main content is as follow:1)For oil-water two-phase flow with high water hold up, propose a Block Multi-threshold method to improve the accuracy of phase fraction. In addition, firstly convert the image reconstruction problem of WMT into solving linear equation, and then based on spectral analysis, propose a new method by modifying singular value spectrum for improving the uniformity of sensing field to increase imaging quality.2)Use TP-GLSQR algorithm which combines Tikhonov preconditioner and general norm to accelerate convergence rate and make phase boundary much clearer. Also, based on inner regularization, propose a new iteration stopping rule to improve the negative effect of the semi-convergence property of Krylov subspace methods.3)Use TwIST algorithm which takes full advantages of the de-noising ability of IST and the effectiveness of solving ill-posed problem of IRS to increase the stability and imaging quality.4)Use MWPCA algorithm to separate pulmonary signal from noise and cardiac signal by performing PCA of the wavelet coefficient matrix of measurement data, which effectively improves the stability of inverse problem algorithms. |
PDF Full Text Request