Segmentation and reconstruction of cultural heritage based on photogrammetric point clouds

Žiga Stopinšek (2016) Segmentation and reconstruction of cultural heritage based on photogrammetric point clouds. Prešeren awards for students.

PDF Download (1528Kb)

Abstract

In this thesis we focus on reconstruction and segmentation of cultural heritage based on photogrammetric point clouds. We present the methodology of archaeological 3D documentation on land and underwater, state-of-the-art reconstruction and segmentation techniques, and prepare an overview of use cases and current issues in computer science in archaeology. We updated Segmentor, a tool for segmentation with superellipsoids, ported it into a library and included it as a plugin in 3D modelling tool Meshlab, a well established tool for 3D modelling. We implemented a supervised method of creating initial seeds for segmentation and a global deformation method for superellipsoids, suitable for modelling amphoras and similar artifacts. We reconstructed a dozen of superellipsoids from 3D models of archaeological sites and of different amphoras accessible on the web.

Actions (login required)