Parallel chaotic iterative algorithms for image reconstruction with limited projection data

Abstract

The parallel chaotic iterative algorithms for image reconstruction by method of asynchronous chaotic relaxation with delay using the Monte-Carlo method are proposed. These algorithms are some generalization of parallel chaotic iteration methods considered by Bru, Elsner and Neumann. The accuracy and the rate of convergence of these algorithms are evaluated through their computer simulation with application to physical researches by tomographic reconstruction from incomplete data. Numerical calculations for solving this problem for some modeling objects, comparing evaluations of errors and the rate of convergence of these algorithms are presented.

These algorithms may be realized effectively in speed independent computing network. This network consists of interacting speed independent processors, which are nonsynchronous devices, and duration of computational processes is defined by duration of transient processes in them. So computing time of each macroiteration for these algorithms in such network is considerably less than computing time for corresponding algorithms in synchronous computing structure.