the description and
of main interest, to
3, organization and
e. The paper shows
use of the data for
1odel, resulting in a
ate frame the local
tity measures and
nto a unique object
lapping areas.
rocess collecting all
c transformations to
ich die vollstándige
on Photogrammetrie
ar zu bestimmenden
hreibung ein echt
wahl, Bestimmung,
wungen geben den
rung erlauben.
den verschiedenen
S ZzZ(X,Y) Funktion
stem lassen sich die
r Punktbestimmung
ammenfassung der
Fortlassen oder die
nerierung sind über
e radiometrisch zu
ce is described as a
Z(X, Y), assuming Z
nage parallaxes.
on satisfies many
3 (eg topographic
extensions in the X
nall range forced by
gy. However, this
le to volume objects,
The discrete surface description (digital elevation/object
models (DEM,DOM)) serves as data base for different
purposes. Common are graphical/visual reproductions
(orthophotos, maps, perspective views) or numerical
analysis. This might be achieved by putting the data into
specific terrain modelling programs or into geographical
information systems (GIS).
However, the use of DOM data within GIS is restricted to
Z=Z(X,Y) surfaces, because of the planar geometry,
supported by the today GIS software. As consequence,
there arise some restrictions in the use of height
informations:
— no z-oriented query within the data possible
— no height ambiguities allowed (eg as existing at
buildings)
— the object has to be a surface above the xy-plane
To overcome these problems new data models have to
be develloped. But the change from the todays 2D+1D
GIS to real 3D GIS is a very demanding way, leading to
complete new data structures, query and analysis tools.
One step into the direction of real 3D GIS is to find a way
for the determination and uniform description of surfaces
from volume objects as they are found in disciplines like
architecture, archaeology or industry.
A contribution to this step will be presented here, resulting
in the complete description and visualization of the
surface of a roman bust.
CONCEPTION
Several aspects have been considered in order to
generate the DOM of the volume surface and their
visualization (cf fig. 6):
use of common algorithmic tools as far as possible
use of digital photogrammetric techniques
provide a homogeneous geometry
minimize the error propagation
provide quality measures for the surface data
make flexible use of the data
allow different forms of presentation
$4.4 4 44 4
The first two and the last aspect are realized using the
following software packages:
photo triangulation : BLUH (Jacobsen, 1980)
image matching ARCOS (Boochs, 1989)
visualization : IDL (Creaso, 1995)
ARC/INFO (ESRI,1994)
The other criterions are met by the following strategy
— definition of a unique object bounded coordinate
system for the description of the whole surface
definition of a circular image configuration
simultaneous triangulation of all images
generation of the DOM by combination of all local
compiled surface models
— comparision and aggregation of multiply determined
surface parts
— development of a data structure allowing the use of
the available quality measures
441
In detail, the workflow for the project looks as shown in
fig.1.
Flowchart showing steps from data capture up to
3D object model and products thereof
planning ot optimal photo recording
image configuration
simultaneous triangulation within a E ehe]
uniform 3D coordinate system Versi
i
selection of and transformatiory | compilation of surface models
into local coordinate systems || by means of image matching
i
backtransformation into
the uniform 3D system
> data selection
f volume object |
aggregation of all
local surface models
Y
generation of a data structure
with full topology and point attributes
. : try
raphical presentation |< geome
"ra
lines densities
i contour lines ) f ortho image ) shaded surface )
(perspective grid)| | animation ) (virtual surface model)
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
Figure 1: Flowchart of the processing steps
REALIZATION
The object. As object for this investigation the marble
head of a statue of the empress Ariadne (AD 474 - 515)
has been chosen. The original found in 1888 is kept in
Rome. For carrying out the work, a copy was put to our
disposal by the Rómisch-Germanisches Zentralmuseum
(Mainz).
The sculpture shows a surface with low contrast and
noncontinuous areas, with only few discrete points
immedeatly appearing useful for the photo triangulation.
Furthermore, problems in lightning have to be considered
due to the smooth shape of the surface. In opposition to
other close range applications, the surface was not
lightned by a texture projector, because the image data
should be used for visualizations based on the natural
shape of the sculpture.
As a close-range-equipment was used, special care has
to be given to the generation of the photos, because of
size and geometry of the sculpture, possibly resulting in a
lack of sharpness within the depth range.
The facts mentioned have made the work quite difficult,
what has been accepted in order to demonstrate the
capability of the method used.
Image configuration. The whole surface has to be
covered by stereo images which should be arranged as
image pairs with parallel camera axis in order to simplify
manual measurements with an analytical plotter. From
the viewpoint of the photo triangulation a more dense
coverage of the object would be useful, what together
with the choice of an optimal geometry for the intersection
of corresponding image rays leads to a symmetric
configuration.
