Accelerated SIRT Algorithm Based on Multigrid

Under the framework of Simultaneous Iterative Reconstruction Technique(SIRT),a new multigrid-based accelerated algorithm is proposed.For each pixel,SIRT updated correction is relevant with pixels which all covered rays go through.In view of this,by using multigrid method,we analyze SIRT and points out its global iterative characteristic.Then inspired by the strategy of "smooth in fine level,correct in coarse level",we achieve the fast SIRT image reconstruction.The new algorithm proposed in this paper aims at reducing computational effort,and does not conflict with other common accelerated algorithms which focus on accelerating the iterative convergence.Numerical experiments show that the new algorithm is better for noise suppression,while the calculation speed in this simulation compared to SIRT is close to double.