-19-
3D-RECONSTRUCTION & RE-PRODUCTION IN ARCHAEOLOGY
M. Ioannides a , A. Wehr 2 3
a Higher Technical Institute, P.O. Box 20423, CY-1521 Nicosia, Cyprus - gammat@cytanet.com.cy
b Institute of Navigation, University of Stuttgart, Breitscheidstr.2, D-70174 Stuttgart, Germany -
wehr@nav.uni-stuttgart.de
KEY WORDS: Imaging 3D-laser Scanning, CW-laser, Scanning Geometry, 3D-digitization, 3D-Reconstruction, point clouds
processing, CAD-modeling, Reverse Engineering, Cultural Heritage Recording
ABSTRACT:
First an imaging 3D-Laser Scanner (3D-LS) will be presented and its functioning will be explained. The scanning mechanism of the
3D-LS is realized by two Galvanometer Scanners. Using such a scanning device it is possible to program various scanning patterns
depending on the application. Therefore, the scanner can be applied for the digitization of small sculptures as well as for digitizing
interiors of rooms. It has been shown that the 3D-LS is an excellent tool for 3D-Reconstruction. However, using laser scanners for
digitizing comprehensively archaeological items means that several hundreds of Megabytes must be efficiently processed. This
problem will be addressed in the second part. It will be shown, how laser scanner data must be processed to obtain NC-programs for
re-production in archaeology. Here the Advanced Surface Modeling Software Package (ASMOS) developed by the University of
Stuttgart and the Higher Technical Institute is applied. Also the problem reconstructing solids by using 3D-LS data from different
overlapping views will be addressed and examples will be presented. Various examples will demonstrate that 3D-LS can be well
applied in 3D-reconstruction and re-production if sophisticated and powerful software packages are available for processing the 3D-
LS image data efficiently.
1. INTRODUCTION
In the last years 3D-Laser Scanners (3D-LS) have been used
more and more in 3D-reconstruction and reproduction, because
very precise 3D-data can be measured in very short time
intervals. Due to the high point density not only 3D-surface
models can be generated but also real 3D surface images can be
derived if intensity data are available.
In the beginning of 3D surface digitization either with
photogrammetric means or tactile machines a modeling of the
surveyed surface was required, because the number of
measurement points was very limited. Today the problem is
vice versa. The user obtains a huge amount of data which very
often contains redundant information. 3D point clouds
containing more than a million measurement points are typical
for laser scanner data sets. In order to reduce superfluous data
again surface models must be generated. This means
sophisticated processing software is necessary to obtain valid
data for CAD modeling or even rapid prototyping. As soon as
CAD data are available 3D-reconstruction can be carried out by
commercially available CAD software. In this processing level
reconstruction of volumes are also possible. Therefore, in this
paper main emphasis will be laid on the modeling
postprocessing algorithms and software respectively. However,
before all sorts of algorithms and procedures are discussed and
analyzed, a 3D-LS developed by the Institute of Navigation will
be explained. The scanning pattern can be customized by
software modifications. Therefore, the 3D-LS covers a large
field of applications as e.g. scanning of sculptures, small objects
and interiors of rooms.
2. 3D-LASER SCANNER
Over the past years the Institute of Navigation of the University
of Stuttgart has developed several imaging 3D laser scanners
using continuous wave (cw) semiconductor lasers as
transmitters and carrying out the slant range measurement by
applying the phase difference measurement principle which is
also known as side-tone ranging. Here the intensity of the light
Figure 1. Principle 3D-LS Setup
of the cw-semiconductor laser is modulated by the drive current
with high frequency signals. As the phase difference between
the transmitted and from the object surface backscattered laser
light is proportional to the two-way slant range depth
information can be directly measured. Using high modulation