Michael Wimmer

(Technische Universität Wien, Institut fuer Computergraphik und Algorithmen, Austria)

Purpose: The accurate reconstruction of high-quality representations from range scanning devices for archaeology is very time consuming and costly. The objective of this paper is to show that this accurate reconstruction step can be avoided in many cases. Instead, we present a method to make range scanning data instantly available to archaeologists and other scientists, so that they can immediately experiment and work with the data.

Methodology/Approach: We discussed with archaeologists, and discovered that one of the main problems – apart from the cost of the devices themselves – with range scanning system in archaeological research is the complexity of creating reconstructions from the scanned data. This tasks requires the intervention of qualified (and expensive) experts, and even then the availability of the reconstructions is usually weeks or months away from the scan. Therefore, our approach was to implement a system that allows users to directly interact with the scanned data.

Results: The system we implemented is capable of visualizing huge range scanner data sets. For example, we took a scan campaign of the Vienna Stephansdom which consists of over 100 different scan positions. The total amount of data is almost 4GB (1 DVD) or 262 Million individual data points. Only 2 hours of automatic processing are required to provide an interactive experience within this dataset, at an update rate of 12 frames per second or higher.
The availability of such a system is very important for archaeologists and other scientists: it gives them a possibility to explore the dataset immediately. They can for example:

  • generate quick visualizations for the public, for investors, or for educational reasons
  • use the model as an indexing structure for annotations, for more detailed models, or for more detailed acquisition planning
  • experiment with placements of other virtual objects, or, after segmentation, rearrange objects within the dataset

Keywords: range scanning, virtual reconstruction, point-based rendering