X-Ray CT Reconstruction Algorithms With Sparse Radiogranphs Based On Bayes Estimates

The essence of CT imaging is computer tomography, widely used in medical, geological exploration and other fields. And it is an indispensable technology for detecting the internal information of the object. At present, the main algorithms of medical CT are direct-projection method, iteration and so on. With the in-depth study, however, appearing sparse data imaging, which traditional methods can not used to reconstruct a clear image.In order to solve the problem of the sparse data reconstruction, this paper, has researched the new methods that based on Bayes estimates for the CT reconstruction.First is to introduce the background of CT imaging, history, current situation, and the current medical CT model, the theoretical basis and mathematics basis for CT reconstruction.Second is in-depth study based on the principle of the CT reconstruction methods which based on Bayes estimates .Then analysis the concrete embodiment of priori information, likelihood function and posterior probability density in the X-ray imaging. On the basis of Bayes theory, use of language Fortran to reconstruct the sparse data model. Adopt structural model and combine with the X-ray attenuation from scanning, to obtain the posterior probability density of images, through Gibbs sampling, using an estimated average as the reconstruction image. The key factors of impacting picture quality are scope of a priori and a posteriori the choice of sampling frequency. Reconstruct the noise model to study the anti-noise capability of Bayes reconstruction methods.Third analysis the principle of algebra method (ART algorithm), and use ART algorithm to reconstruct image. The key factors of impacting picture quality are the number of block and the number of iteration. Finally, used the CT reconstruction algorithms based on Bayes estimates and ART algorithms to reconstruct image, and compared each other from image-resolution, real-time, anti-noise capability and so on . On this basis, combined the advantages of the two methods, and used a result of small block and less iteration as the prior information of the CT reconstruction algorithms based on Bayes estimates when there has on prior cases,get a good reconstruction image.