Abstract
This paper deals with a theoretical possibility of a new visualizing measurement method based on a fast 2D model reconstruction utilizing a few projection data. A theory of least squares syped 2D model reconstruction by means of cutting singular value decomposition is discussed, utilizing projection data of a few directions for a smooth 2D spampled density distribution model which satisfies the condition of the sampling theorem. First it is shown that we can set up a linear equation system which corresponds to the parallel penetrating X beams. Next in advance by means of singular value decomposition and cutting small singular value down, we get a constant matrix for the linear equation system. Last we can fast calculate a good 2D image with the least squares error in the reconstruction by only one multiplication of matrix. The results of computer simulation with the fast 2D reconstruction algorithm are presented. Key words : Fast 2D Model Reconstruction, Sampling Theorem, Smooth 2D Sampled Density Distribution Model, Cutting Singular Value Decomposition
