imbul 2004
. that is to
nents and
t has been
ent. Using
'onnectors
of objects.
lex object
instantiate
roduction
different
natch the
the local
facilitated
gorithms
arate. It is
s specific
matically
.. This ap-
| even for
automatic
lanes and
red. The
1ise and
the points
very dark
tained by
1 complex
takes the
sh which
of points.
verall size
has been
asis.
station SE
3eometry)
aser Gen,
open to
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3
2.2 The DLR camera Eye Scan
The next section introduces the main features of the DLR line
scan camera.
SYSTEM FEATURES
This camera Eye Scan (see Figure 7) is developed in a common
project between an industrial company and the German
Aerospace Center (DLR) for environmental documentation pur-
poses. The camera consists of a rotating unit which rotates an
integrated CCD line chip 360? to achieve a full view of the
environment [Sch04]. By using three line chips with 10,000 ele-
ments each a very high resolution can be achieved. Three lines
(RGB) provide each 16 Bit information of the environment and
guarantee a high dynamic range with monitoring.
Figure 7: The figure shows the high resolution DLR line scan
camera.
The resulting images consist of a maximum of 10200 by 500000
pixels each containing three 14bit RGB values. The image is
stored by a specially developed frame grabber onto the hard
disk of a computer. A typical scan (10000 x 30000) using a
special optical lens system by 35mm optical focus length takes
about 3 min (daylight) and up to 60 min (dark indoor illumi-
nation), mainly depending on the ambient illumination
conditions and the number of rows to be measured with the
camera.
SOFTWARE
The software for the camera enables the user to adjust typical
camera settings like e.g. the shutter speed. For improving the
homogeneity of the colours for different illumination
conditions, it is possible to choose shading correction tables
(e.g. daylight or indoor illumination). The software also
includes a package for the geometric calibration, which cnables
a recalculation of the rawdata into a geometrically calibrated
image.
In the next paragraph it will be described how 2D RGB color
information can be mapped onto a 3D laser point cloud.
2.3 Combining camera and laser scanner data
In the photogrammetric field, a couple of techniques are known
to map RGB data onto range data: Assuming that the distortion
of both systems is already recalibrated geometrically. the overall
mapping formula must in some way contain a translation (3
unknowns). rotation (3 unknowns) and perspective projection:
In the following it will be shown how this 6-dimensional
parameter space can be reduced to one parameter for the given
set-up.
. Istanbul 2004
SETUP
The scanner and camera acquisitions are. performed one after
the other: After finishing a scan with the IMAGER 5003, the
DLR Eye Scan camera is set onto the same tripod. by fixing an
adapter onto it (see Figure 8). This adapter ensures that the
location of the optical centre of the camera is (nearly) identical
with the one of the scanner unit, as well as both horizontal
rotation axis are. Thus the only unknown parameter, which must
be calculated to transform both co-ordinate systems into each
other, is the horizontal angle.
Figure 8: The figure shows the basic principle of the data
acquisition (see text).
3. APPLICATIONS
ARCHITECTURE AND HERITAGE
For a lot of applications in the field of architecture and cultural
heritage not only geometry is enough but also reflectance and if
possible colour information is necessary to fulfil requirements
of architects.
The following chapter presents examples using both systems:
3.1 Castle Neuschwanstein
In this example the results of a project between Z+F and the
DLR are presented. The task was to scan and survey the inside
of the drawnhall and create a 3D colour model.
shows the drawnhall of castle
Figure 9: The figure
Neuschwanstein: The reflectance image of the
Imager is overlapped with the corresponding RGB
colour information of the Eye Scan camera.
