Binary Tomography by Iterating Linear Programs from Noisy Projections [chapter]

Stefan Weber, Thomas Schüle, Joachim Hornegger, Christoph Schnörr
2004 Lecture Notes in Computer Science  
In this paper we improve the behavior of a reconstruction algorithm for binary tomography in the presence of noise. This algorithm which has recently been published is derived from a primal-dual subgradient method leading to a sequence of linear programs. The objective function contains a smoothness prior that favors spatially homogeneous solutions and a concave functional gradually enforcing binary solutions. We complement the objective function with a term to cope with noisy projections and evaluate its performance.
doi:10.1007/978-3-540-30503-3_3 fatcat:dbg2rfeld5f7tjdzh6ir36bn3m