Total variation-based neutron computed tomography.

摘要: We perform the neutron computed tomography reconstruction problem via an inverse formulation with a total variation penalty. In case of highly under-resolved angular measurements, penalty suppresses high-frequency artifacts which appear in filtered back projections. order to efficiently compute solutions for this problem, we implement split Bregman algorithm; due error-forgetting nature algorithm, computational cost updating can be significantly reduced very inexact approximate linear solvers. present effectiveness algorithm low-angular sampling using synthetic test problems as well data obtained from high flux source. The removes and even roughly capture small features when extremely low number angles are used.