Research On Exciting-measuring Modes And Image Reconstruction Algorithms For Electrical Tomography

Electrical tomography is one kind of nondestructive and visible measurement technique. Electrical tomography has wide applications in medical and industrial fields for the advantages of being fast in response, nonintrusive and capable of acquiring 2D/3D image information of distribution parameter.Successful applications of electrical tomography technique depend on the accuracy of image reconstruction algorithm which is determined by the'soft-field'characteristic of sensing field and limited projection data. In order to improve the accuracy of reconstructed images, the above two aspects are deeply studied in this paper. The main work and conclusions are as follows:1. As an electrical impedance tomography system used in clinic monitoring, the spatial resolution of reconstructed images is low for the complicated body structure. Based on the fact that the body conductivity distribution varies slowly, Landweber iterative algorithm with updated sensitivity matrix is studied to improve the quality of reconstructed images. The initial image for sensitivity matrix update is obtained by Landweber iterative algorithm, and the results of sensitivity matrix update using different initial images are compared. The times of sensitivity matrix updating are also analyzed. Experimental and simulation results show that reconstructed images with higher accuracy can be obtained.2. Driven pattern is the way to inject current or voltage at the boundary of an object. When adjacent current driven pattern of EIT is used, the currents concentratedly distribute in the area adjacent to the electrodes compared with the central area of pipe, which causes the low quality reconstruction images. For a 16-electrode EIT system, the uniformed adjacent voltage measurement mode is adopted and the eight current driven patterns are investigated considering the structural symmetry. The isopotential lines, the number of independent measurements, measured voltage dynamic range, voltage change with conductivity change and reconstructed image quality of these patterns are compared, through which the seventh pattern is selected.3. The capacitance normalization models of electrical capacitance tomography are investigated in detail. The calculation method of sensitivity matrix is proposed, which is based on the combined model and the distribution of electrical field lines. The Landweber iterative algorithm with optimal step length for the combined model is derived. Simulation and experimental results show that objects can be clearly distinguished from the reconstruction images with distinct edges and good fidelity using the proposed algorithm.4. Electrical tomography is a typical nonlinear mapping problem. Wavelet neural networks combine the localization characteristic of wavelet and self-learning ability of neural network. It has good function approximation and error tolerance capabilities, together with fast convergence speed. The image reconstruction algorithm based on wavelet neural networks for electrical capacitance tomography is proposed. Principal component analysis method is adopted to reduce the dimensionality of input data. Simulation results show that reconstructed images with high accuracy can be obtained and the computational speed is near to that of linear back-projection algorithm.5. In order to increase the projection data of electrical tomography system, electrical capacitance tomography with combined electrodes is studied. Excitation and measurement modes based on combined-electrode strategy of electrical capacitance tomography are proposed. Different combined-electrode strategies are investigated, which can effectively increase the projection data of electrical capacitance tomography. The combined-electrode excitation and measurement strategy need not to increase much more hardware costs, which makes it flexible to realization. Simulation results show that the accuracy of reconstructed images can be enhanced clearly for the combined-electrode electrical capacitance tomography.