Transmission And Scattering Tomography Inversion Algorithm

The technique of computerized tomography is not only a milestone in the history of radiodiagnosis but also important means for nondestructive testing in industry .Filter Backprojection algorithm and Algebraic Reconstruction Techniques are commonly used in CT : This thesis lucubrate these algorithm , present pivotal and innovative amelioration ,improve their speed and accuracy .Filter function is a key for the Filter Backprojection in CT . In this paper a new filter function is suggested ,its primary part is a smooth operator in functional analysis.The new filter has many excellent specialty .and the error estimation of CT due to use it is given. We note that the Fourier transform of a function with compact support is not compact , but its side-peaks fast decay.The new filter even has this specialty ,so local tomography can be implemented .In this thesis there is a principle of a kind of local tomography algorithm with the new filter. The second error estimation is gained. The global and local reconstruction of simulation and experimental projection data show that by the local tomography algorithm resolution and Gibbs effects of the reconstructed images can be improved This method of local tomography is very simple and fast Their reconstruction time is approximately the same as proportion of their reconstruction region ,when the local reconstruction region is a tenth part of the global . The effect of the local reconstructed image isn't more bad then global.On algebraic reconstruction techniques for image reconstruction, there are great difficulties in being fast reconstructed due to heavy data and great memory to be dealt with , in which one of important aspects is how to seek for the nonzero coefficients of projection matrix. In this thesis a principle of geometrical symmetry for image reconstruction is discovered ,the calculation of the nonzero coefficients of projection matrix has been simplified by using this symmetrical principle.This principle doesn't concern with distributing of the object function therefore it can be used in any reconstructed problem .By using this symmetrical principle, a symmetric block-iterative method is investigated for the image reconstruction problem. One also demonstrate the efficiency of our implementation of symmetric algorithm with simulation and experimental projection data and compare its performance with respect to other reconstruction techniques, and then analyse several calculated measures of merit. The results show that the algorithm presented here is more fast and has higher resolution factor than other.Comparing with the transmissin tomography, the Compton scattering imaging technique has some superiorities such as high sensitivity, more freedom in the arrangement of experimental setup and can be used directly for 3-D imaging. However, this technique is stifled by difficulties such as the nonlinear and ill-posedness problem due to the attenuation of incident and scattering rays and the multiple scattering problem. Here, we will focus on the first problem.Commonly the transmissin tomography and the scattering imaging technique are researched respectively, but yet analyzed relatively according as the imaging systemin our work. By analyzing the whole physical process of Compton scattering, we have educed newly the analytic reconstruction expression of the backscatter imaging putted by Norton, which has a physical meaning in accordance with its definition.A principle of the line-scan Compton backscatter imaging based on scattered photon energy spectra is presented. This imaging equation is concluded as a large-scale system of nonlinear equations. In view of the neural network with multi-criteria optimization, a solution of image reconstruction is worked out accordingly. The results of data simulations illustrate the fact that it is feasible to adopt the method of constructed neural network for the sake of multi-criteria Compton backscatter imaging, which is applicable to solving the ill-posed reconstruction problems.Moreover,in this thesis a new...