Filtered stochastic optimization for binary tomography

In this work, we use stochastic diffusion equations to improve the reconstruction of binary tomography cross-sections obtained from a small number of projections. A first reconstruction image is obtained with the Total Variation regularization method. The reconstruction is then refined with stochastic approaches. This method is applied to a noisy bone cross-section with 10 projection angles. The main purpose of this work is to show the improvements obtained with a filter taking into account wavefront set properties of the Radon transform.