Generalized row-action methods for tomographic imaging (Q742852)

The tomographic image reconstruction is solved by row-action methods. The used methods can be viewed as incremental gradient methods, that means a process which minimizes the sum of a large number of convex functions. In some cases the rate of convergence of the method is very poor, but it could be acceptable. The paper develops relaxed variants of a class of incremental proximal gradient methods, which, as it follows, generalizes known row-action methods. The new incremental algorithms for tomographic imaging are derived. The efficiency of the method is demonstrated by numerical examples.