Towards Surface Regularization via Medial Axis Transitions

Originality: Presents the first use of 3D medial graph transitions to structurally simplify surfaces, i.e., modify their structure based on their regional features (corners, ridges, …) without modifying other features. This approach to simplifying shapes is more controllable than previous methods in that the effects are more local. Rigour: Tested on artificially deformed data to prove the successful recovery and then on real data. Significance: The algorithm is used in a large NSF-funded archaeology project at Brown University . ICPR is the major international conference in pattern Recognition. Peer-reviewed by 3 with an 18% acceptance rate for full papers.