Employing Graph Cut for qualitative Volume Reconstruction

This paper presents a 3D volume reconstruction algorithm. The input of this algorithm is made up of only two orthogonal 2D projections. Owing to this extremely small input set, the problem becomes highly ill-posed. Our algorithm uses the graph cut techniques in order to efficiently solve this problem. Our method does not assume any hypothesis about the unknown volume. This presented algorithm uses only information extracted from projections to define all the graph cut costs. We present a example of prior information which could be used to improve the results. This paper presents some real complex results reconstructed volumes. It also contains a comparison with an other volume reconstruction method method and a brief discussion of the most important terms.