Geometry Based Faceting of 3D Digitized Archaeological Fragments

Hanan ElNaghy, Leo Dorst; Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2017, pp. 2934-2942

We present a robust pipeline for segmenting digital cultural heritage fragments into distinct facets, with few tunable yet archaeologically meaningful parameters. Given a terracotta broken artifact, digitally scanned in the form of irregularly sampled 3D mesh, our method first estimates the local angles of fractures by applying weighted eigenanalysis of the local neighborhoods. Using 3D fit of a quadratic polynomial, we estimate the directional derivative of the angle function along the maximum bending direction for accurate localization of the fracture lines across the mesh. Then, the salient fracture lines are detected and incidental possible gaps between them are closed in order to extract a set of closed facets. Finally, the facets are categorized into fracture and skin. The method is tested on two different datasets of the GRAVITATE project.