Research on CT iterative reconstruction algorithm based on noise model and image constraint

This research aims to reduce the effect of noise on reconstructed image of CT. We give a method which adds noise model and image constraint to the CT iterative reconstruction algorithm. The image constraint we use is weighted Total Variation minimization method. The noise model depends on the types of noise. On one hand, the noise in the projection data before logarithmic transformation is approximated by the Poisson distribution. One the other hand, the noise in the projection data after logarithmic transformation is approximated by the Gaussian distribution. Finally, we use the Split-Bergman method to solve our final objective function. In our numerical experiments, we adopt different methods to dispose the projection data in which the noise is already added. And by comparison between the results, we test feasibility of our method. In conclusion, the completed algorithm which is presented in this research is better than other methods. And it can effectively reduce the noise effect and improve the image quality. In the future, we will optimize the algorithm which obtained in our research and extend the methods for 3D construction and apply it for multi-source tomography.