This thesis presents a framework for image segmentation and 3D reconstruction with topological constraints, a problem class not yet solved sufficiently by existing methods. The proposed method is specifically successful, when the object has a thin shape, e.g. when reconstructing vascular networks in medical image analysis, and for 3D reconstruction of thin structures. The constraints can be formulated as linear constraints in a convex optimization framework, which allows for a globally optimal solution.
«
This thesis presents a framework for image segmentation and 3D reconstruction with topological constraints, a problem class not yet solved sufficiently by existing methods. The proposed method is specifically successful, when the object has a thin shape, e.g. when reconstructing vascular networks in medical image analysis, and for 3D reconstruction of thin structures. The constraints can be formulated as linear constraints in a convex optimization framework, which allows for a globally optimal s...
»