Euclidean skeletons using closest points
Songting Luo, Leonidas Guibas, Hongkai Zhao · Inverse Problems and Imaging · 2011
In this paper, we present an efficient algorithm for computing theEuclidean skeleton of an object directly from a point cloudrepresentation on an underlying grid. The key point of thisalgorithm is to identify those grid points that are (approximately)on the skeleton using the closest point information of a grid point and its neighbors. The three main ingredientsof the algorithm are: (1) computing closest point informationefficiently on a grid, (2) identifying possible skeletal pointsbased on the number of closest points of a grid point and itsneighbors with smaller distances, (3) applying a distance orderedhomotopic thinning process to remove the non-skeletal points whilepreserving the end points or the edge points of the skeleton.Computational examples in 2D and 3D are presented.