A new ab initio reconstruction method from unknown-direction projections of 2D binary set

This article focuses on the tomographic reconstruction of 2D binary images from projections whose directions are unknown. Observing that there is a mutual influence between the image reconstruction and the directions reconstruction, a new ab-initio method based on joint reconstruction of directions and image is proposed for 2D binary objects. Its implementation is done by optimizing a cost function taking as arguments the input projections, the current reconstructed image and the current directions. The optimization of the cost function is based on simulated annealing. After detailing the cost function, we test the proposed method for different resolutions and different noise levels. Finally, we present the future development of this promising method.