A surface approximation algorithm for complex shapes

In the paper, we propose one surface approximation algorithm which could handle the geometrical and topological errors contained by the complex shapes. The algorithm firstly builds one initial coarse envelop based on the voxel hulls of the input shape, then one iterative approximation is processed by two alternative steps: matching the envelop to the input shape by using the Laplacian deformation, and the Laplacian fairing. The experiments demonstrate that our algorithm could efficiently build one surface that approximates the input shape, which could be in by arbitrary representations, such as meshes, polygon soups, and point sets.