Superiorized regularization of inverse problems

Abstract

Inverse problems are characterized by their inherent non-uniqueness and sensitivity with respect to data perturbations. Their stable solution requires the application of regularization methods including variational and iterative regularization methods. Superiorization is a heuristic approach that can steer basic iterative algorithms to have small value of a certain regularization functional while keeping the algorithms simplicity and computational efforts, and is able to account for additional prior information. In this note, we combine the superiorization methodology with iterative regularization and show that the superiorized version of the scheme yields again a regularization method, however accounting for different prior information.