Optimising 3D Mesh Unfoldings with Additional Gluetabs using Simulated Annealing

Thorsten Korpitsch

Supervisor(s): Hsiang-Yun Wu

TU Wien

Abstract: 3D Mesh unfolding is a process of transforming a 3D mesh into one or several 2D planar patches. The technique is widely used to produce papercraft models, where 3D objects can be reconstructed from printed paper or paper-like materials. Nonetheless, the reconstruction of such models can be arduous. In this paper, we aim to unfold a 3D mesh into a single 2D patch and introduce Gluetabs as additional indicators and in order to give users extra space to apply glue for better reconstruction quality. To avoid unnecessary Gluetabs, we reduce their number, while still guaranteeing the stability of the constructed model. To achieve this, a minimum spanning tree (MST) is used to describe possible unfoldings, whereas simulated annealing optimisation is used to find an optimal unfolding without overlaps. We aim to unfold 3D triangular meshes into single 2D patches without applying shape distortions, while appropriately assigning a reasonable amount of Gluetabs. Moreover, we incorporate a visual indicator scheme as a post-process to guide users during the model reconstruction process. Our quantitative evaluation suggests that the proposed approach produces fast results for meshes under 400 faces.
Keywords: Computational Geometry, Geometry Processing, Human-Computer Interaction
Full text:
Year: 2020