BLOCK-ITERATIVE ALGORITHMS FOR IMAGE RECONSTRUCTION

M Jiang,G Wang
DOI: https://doi.org/10.1142/9789812777171_0006
2005-01-01
Abstract:In this paper, we first introduce a general iterative scheme for image reconstruction and then apply it to CT image reconstruction with practical data. This scheme can be readily formulated in the simultaneous and block-iterative formats. This provides a facility to obtain new iterative algorithms from known methods and also permits more flexibility in task scheduling for parallel implementation. It can be shown that the simultaneous algebraic reconstruction technique (SART), is a special example of the general scheme. Based on the convergence properties of the general scheme under mild conditions on the relaxation parameters, in the consistent or inconsistent cases, the convergence of the SART and its block-iterative version can be easily derived. The block-iterative SART is applied to reconstruct CT image from practical data to demonstrate its performance. Some issues related to the parallel implementation are also addressed.
What problem does this paper attempt to address?