Abstract
Abstract
Symmetry in a mesh can be described as invariancy in specific transformations, such
as reflection, translation and rotation. In this study, we propose a new approach for
extracting the intrinsic curve that separates the mesh into two, each of these being the
reflective symmetry of other.
The algorithm starts with sampling points around the mesh, using AGD (Average
Gedoesic Distance) , followed by MGD (Minimum Geoedsic Distance) and acquiring
the local extremums in both. Our algorithm matches sampled points using HKS and
their respected distances to the mesh’s midpoint. The matching uses the bins that are
generated from candidate voronoi regions and their geodesic paths to the candidate
points. Following that, the algorithm detects and prunes the best suited voronoi region
generated between matched points. With the voronoi region used as symmetry curve,
we can infer new and stronger matches from sampled points with new iterations.