|M.Sc Student||Zamansky Alik|
|Subject||A Framework for Surface Reconstruction of Sparsely-|
|Department||Department of Computer Science||Supervisor||Professor Gill Barequet|
|Full Thesis text|
The problem of reconstructing a three-dimensional object from a set of planar parallel slices has been the subject of study for many years. In this problem, each slice consists of simple closed polygonal contours, and contours from adjacent slices should be connected by a surface to reconstruct the original spatial object (so that it can later be visualized and investigated). Several algorithms have been proposed for the reconstruction problem when the distance between every two successive slices is small. However, when the original object is not sampled densely enough, the problem becomes more complex, and the existing algorithms often fail to produce satisfactory results.
In this thesis, a framework is proposed for solving the surface-reconstruction problem for sparsely-sampled objects. The framework consists of several phases which solve different tasks of the reconstruction: obtaining correspondence between contours, resolving branching situations, and producing the actual surface. Different algorithms can be integrated in the framework to construct a correct (non-self-intersecting) surface. Instances of such algorithms are provided as a case study for the framework, and are implemented for testing and evaluation of the proposed method. Experimental results show improvement in quality of the reconstructed surface for sparse inputs, as demonstrated by a few synthetic and real examples.