Skeleton is an important 1-D descriptor of polygon and a useful tool for advanced geometric algorithms. However the skeleton that most existing algorithms investigated is the longest one, which is not intended in all circumstances. This paper proposes an algorithm for extracting hierarchically optimal skeleton network of polygons. The algorithm incorporates length, angularity and area of associated part of skeleton segment. The result is a hierarchical structure and each level corresponds to a specific detail of skeleton. The new algorithm has three steps. First the constrained Delaunay triangulation of polygons is constructed. Secondly skeleton segments are connected between neighbouring triangles and skeleton network is built. Thirdly, a dynamic pruning process considering the weights is employed to produce optimal skeletons at each level of detail. The weight in last step is determined by length, angularity and position of skeleton segment. * Corresponding Author. mapwang@tom.com
[1]
Proceedings first international symposium on spatial data handling
,
1986
.
[2]
Victor J. D. Tsai,et al.
Delaunay Triangulations in TIN Creation: An Overview and a Linear-Time Algorithm
,
1993,
Int. J. Geogr. Inf. Sci..
[3]
D. Lee,et al.
Skeletonization via Distance Maps and Level Sets
,
1995
.
[4]
Olaf Kübler,et al.
Hierarchic Voronoi skeletons
,
1995,
Pattern Recognit..
[5]
Tinghua Ai,et al.
GAP-Tree Extensions Based on Skeletons
,
2002
.
[6]
Borut Zalik,et al.
Sweep‐line algorithm for constrained Delaunay triangulation
,
2008,
Int. J. Geogr. Inf. Sci..