Hierarchical conditionally Gaussian priors, combined with efficient iterative solvers of linear systems, provide an efficient tool for solving inverse imaging problems in which the unknown is believed to be a blocky image. In this talk, we discuss how these methods can be used to guide targeted mesh refinement in imaging problems involving finite element computations.

This presentation is part of Contributed Presentation “CP11 - Contributed session 11”

Authors:

Anna Cosmo (Politecnico di Milano)

Daniela Calvetti (Case Western Reserve University)

Erkki Somersalo (Case Western Reserve University)

Simona Perotto (MOX, Politecnico di Milano)

Keywords:

bayesian methods, inverse problems, numerical linear algebra, statistical inverse estimation methods