by Julian Stoettinger
Abstract:
On archaeological sites, thousands of fragments of ceramics are found. The classification of these findings is very time consuming and still done manually until today. To speed up this process, a 3d scanner for the data acquisition and software for the profile line generation is used. This work provides an end-user friendly user interface for profile line generation of archaeological findings. For performance reasons, the visualization is done by the GPU using OpenGL. 3d scan data with several hundreds of thousands of vertices can be processed in real time. The profile line generation is implemented in three levels: First, profile line generation is simulated by using OpenGL clipping planes and thus, the GPU only. Then, the nearest vertices of the intersection plane are marked and the intersection vertices are estimated. Finally, the profile segments of this diminished 3d scan data are decided. Comparisons to a Matlab implementation and a commercial software have been carried out and these experiments with synthetic and real data are shown. Finally, an outlook towards future development is given.
Reference:
High Performance Profile Line Generation and Visualization using OpenGL (Julian Stoettinger), Technical report, PRIP, TU Wien, 2007.
Bibtex Entry:
@TechReport{TR108,
author = "Julian Stoettinger",
title = "High Performance Profile Line Generation and Visualization using OpenGL",
institution = "PRIP, TU Wien",
number = "PRIP-TR-108",
year = "2007",
url = "https://www.prip.tuwien.ac.at/pripfiles/trs/tr108.pdf",
abstract = "On archaeological sites, thousands of fragments of ceramics are found.
The classification of these findings is very time consuming and still done manually until today.
To speed up this process, a 3d scanner for the data acquisition and software for the profile line generation is used.
This work provides an end-user friendly user interface for profile line generation of archaeological findings.
For performance reasons, the visualization is done by the GPU using OpenGL.
3d scan data with several hundreds of thousands of vertices can be processed in real time.
The profile line generation is implemented in three levels:
First, profile line generation is simulated by using OpenGL clipping planes and thus, the GPU only.
Then, the nearest vertices of the intersection plane are marked and the intersection vertices are estimated.
Finally, the profile segments of this diminished 3d scan data are decided.
Comparisons to a Matlab implementation and a commercial software have been carried out
and these experiments with synthetic and real data are shown. Finally, an outlook towards future development is given.",
}