[Back]


Talks and Poster Presentations (without Proceedings-Entry):

A. Djuricic, P. Dorninger, S. Rasztovits, C. Nothegger, M. Harzhauser, O. Mandic, P. Glira, N. Pfeifer:
"21st International Conference on Cultural Heritage and NEW Technologies";
Talk: 21st International Conference on Cultural Heritage and NEW Technologies (CHNT), Wien; 2016-11-16 - 2016-11-19.



English abstract:
The protected fossil oyster reef in Stetten, Austria is the world´s largest excavated fossil oyster reef, formed by large sea shells. About 50.000 up to 60-cm-long shells cover a 459 m2 large area. The reef consists primarily of Crassostrea gryphoides shells. In this study, our motivation is to reconstruct the original shell positions with automatic 3D object matching and by finding similar or identical objects in a database with determined shell size. Four initial criteria were defined for the object matching: i) iterative neighbourhood search near the examined shell, ii) specified shell convexity: down, up, iii) specified shell side: left, right and iv) shell length with 20% tolerance. For all shells matching the criteria, their centrelines were analysed in the next step. In analysis, the centreline and its neighbouring points are profiled. The profiling produces spatial features, such as sphericity, planarity, scattering and change of curvature. The features describe if the lateral surface of a shell is flat, concave, or convex. All analysed shells are compared to see if they match together by studying left-sided shells with the right ones. The analysis assumes that shell features should be invariant within a potential pair. Finally, the potential matching candidates are brought close together and pairing is completed using an iterative closest point (ICP) algorithm with a constraint that the matching surface cannot intersect between left and right valve.
The proposed method gives a possibility to match and link spatially separated complex objects together if their surface properties have enough feature correspondences along their centreline profiles. The matching over distance supports in making spatial interpretations and objects visualizations in several disciplines, including geology, palaeontology, and biology.

Keywords:
pair matching, fossil shells, center line, automation

Created from the Publication Database of the Vienna University of Technology.