463
PP
1999.
from
'5 Vol
’.ment,
and
DEM
on to
/sical
581-
itions
and
•kshop
and
loniki.
1999.
omplex
'11 Part
object
iustry ",
ction in
irt B5.
iccurate
surface
491.
ibration
CIPA
•01.
project,
icipants
unhofer
aloniki,
Gmbh
Athens,
fuseum
FAST ON-SITE RECONSTRUCTION AND VISUALIZATION OF ARCHAEOLOGICAL FINDS
Konrad Schindler, Markus Grabner, Franz Leberl
Computer Graphics and Vision
Graz University of Technology
Inffeldgasse 16, A-8010 Graz, Austria
{schindl,grabner,leberl}@icg.tu-graz.ac.at
KEY WORDS: Archaeology, Excavation, Close-Range Photogrammetry, Automatic Reconstruction, Interactive Visual
ization
ABSTRACT
We report on a study carried out within the 3D MURALE project, in which the complete workflow necessary for automatic
recording, reconstruction and interactive visualization of artefacts on archaeological excavation sites has been tested. With
state-of-the-art technologies from digital photogrammetry, computer vision and computer graphics a prototypical pipeline
for recording, 3D modeling and visualization of archaeological artefacts has been implemented and tried out on site on
the excavation of Sagalassos (Turkey). The purpose of the presented work was to test the complete pipe-line in a real
archaeologic environment and assess its usability.
1 INTRODUCTION
To meet the practical requirements of archaeological her
itage recording and documentation, technology must be us
able under field conditions on archaeological excavation
sites. Besides being fast and easy to use, systems for pho-
togrammetric recording and modeling as well as visualiza
tion should interfere as little as possible with the excava
tion work and use only equipment that can be found on
archaeological excavations.
We present a system which has been designed for the pur
pose of fast on-site recording, reconstruction and local as
well as remote visualization of archaeological finds. The
system combines methods from photogrammetry and com
puter vision with state-of-the-art computer graphics algo
rithms to accomplish the task. Objects are recorded di
rectly on site with an off-the-shelf digital camera and are
reconstructed in an almost fully automated photogrammet-
ric modeling process, in which the only user interaction
is to segment the artefact to be modeled from the back
ground. Robust algorithms for orientation, adjustment and
dense object reconstruction are applied, which deliver ac
curate, textured 3D object models.
New methods for compressed adaptive multi-resolution en
coding of large models are used to enable visualization of
the resulting 3D models at interactive frame-rates even on
computers with inexpensive graphics hardware. An ex
tension of this technology allows smooth remote interac
tion with the models via low-bandwidth connection using a
standard Internet browser. This gives archaeologists in the
field the possibility to instantly communicate about new
finds with people who are not present on site. In contrast
to existing simplification techniques object identities and
attributes are preserved through the dynamic simplification
process to provide a practically useful way of interaction.
Again the processing is fully automatic.
The proposed system has been successfully tested within
the 3D MURALE project (Cosmas et al., 2001) on the
Greco-Raman excavation site of Sagalassos (Turkey) dur
ing the 2002 excavation campaign. Several artefacts have
been recorded without any impact on the excavation rou
tine, reconstructed on the same day using a standard laptop
computer, and visualized on a different computer via a lo
cal network connection. We will present the utilized work-
flow and technology with real examples from Sagalassos.
2 RECORDING
A classical photogrammetric approach is used for on-site
recording. Multiple overlapping images of the object to be
modeled are recorded with a calibrated digital consumer
camera. In our experiments we have used a high-quality
SLR camera as well as a cheap consumer camera (a Canon
D30 with a resolution of 2160x 1440 Pixels and a Nikon
Coolpix 995 with 2048 x 1536 Pixels). The standard setup
with fixed focus close to infinity has proven to yield sharp
images in most practical situations. We therefore use it,
because it provides the best accuracy and robustness. It
should be mentioned, however, that within the 3D MU
RALE project it has been shown that recording and re
construction on archaeological sites is also possible with
a completely uncalibrated digital video camera (Pollefeys
et al., 2001).
2.1 Camera Calibration
Camera calibration is fully automatic. The user takes 4-6
pictures of a planar calibration target from arbitrary, differ
ent positions. The calibration target consists of a grid of
circles with known positions and has a unique, easily de
tectable marker (a black ring) in the center, which allows to
aoutomatically establish the correspondence between im
age points and target points. The calibration target is de
picted in figure 1. With a calibration algorithm optimized
for this configuration (Heikkila, 2000) a full set of camera
parameters is recovered, i.e. the principal point, the focal
length, two coefficients for a symmetric d^-order radial
distortion polynome and two coefficients for a similar tan
gential distortion polynome. Calibration is an offline pro
cess, which needs to be carried out only once for a given,
stable camera geometry, i.e. it is sufficient to calibrate a