Branch and bound discrete tomography reconstruction of binary contours

In this paper a new method is shown for reconstructing binary contours from its two orthogonal projections. The paper focuses on the reconstruction of contours whose inner territory is hv-convex. Since the inner territory of the contour is hv-convex, the first and the last row and column of ones are comprised from one part, while all other rows and columns consist of two streaks of ones. An additional constraint is that all ones in the matrix have exactly two 4-neighbours. To find the solution, a branch and bound method is used which allows the cutting of the edges during tree development. The algorithm works best for smaller matrices up to size 50×50 pixels.