A Regularization Method for Computed Tomographic Reconstruction

摘要

The problems arising in the computed tomography-area are well known for their high dimensions and illposedness.Tikhonov regularization method is used to reconstruct parameter distribution from projection data.The linear programming and conjugate gradient method are used to compute the regularized solution for the least-square equations.In numerical simulation,the regularization method with linear programming was unable to reconstruct distributions effectively,while the approach based on conjugate gradient method produced reliable asymmetrical reconstructions by computing underdetermined equations and overdetermined equations respectively.The average errors using conjugate gradient regularization method were 2% and the maximum value errors were 5% after 10 iterations,which provided a good indication of the precision and convergence of the method.