—
© LO 0 © J -
—
—
N «D = «cu 2
r
f
al
n
it
transferable to the commonly used format within the
company. In a feasibility study, the following methods have
been considered or tested.
2.2.1 Terrestrial surveying methods with use of existing
maps
The planimetric information can be evaluated by means of
digitizing or scanning of appropriate maps. The vertical
dimension can be achieved by using official information
(number of floors) or hand-held laser devices on the spot.
The up-to-dateness, accuracy and the degree of
specification depends on the used maps; the vertical
accuracy is low (in the case of using floor heights) or
relatively expensive to derive (in case of using surveying
methods or hand-held laser devices). The generation of
additional land use information is possible. The method is
operationally available, the high effort of manual works
leads to higher costs, and controlling and updating of the
delivered data is easily possible.
2.2.2 Analytical photogrammetry
Both the horizontal and the vertical determination can be
done by means of analogue aerial images with appropriate
scales within analytical photogrammetric devices. Further
processing of the derived vector information will be done
with geographical information systems.
The actuality, accuracy and the degree of detail depends
on the date of flight and the used image scale. Both terrain
elevation and buildings heights can be derived in common
processes, additional information can be derived also by
the operator. The method is operational, but due to manual
works costly, controlling and updating (with newer images)
is easily possible.
Alternatively just the heights can be derived by analytical
processing and the planimetry by maps.
2.2.3 Digital photogrammetry
On the base of scanned aerial images (or airborne scanner
images) the horizontal and vertical determination can be
performed by computers. The methods are not yet
operational and differ from the pure transfer of the analytical
world into digital to almost fully automatic procedures.
Automatic 3D recognition is a wide field of research (e.g.
Huertas and Nevatia, 1988; Dang et al., 1994, Haala, 1995,
Lang, Schickler, 1993). Especially in dense urban areas
with complicated building shapes and hidden areas due to
shadow or nature ground cover, automatic processes are
not able to detect the outlines and the heights of the
building precisely. In comparison to analytical methods the
digital approach is cheaper due to automation. However
the accuracy of the human operator will not be reached
(but is it necessary ?) and the costs rise, the more manual
corrections are necessary. Controlling and updating is
easily possible.
2.2.4. Further methods
There are some possible methods which have some
restrictions up to now, but could be valuable in the future.
- ATKIS data (Authoritative Topographic and
Cartographic Information System)
799
the medium - scale mapping database DLM25/1 of
Germany 's federal states support some area-based
objects, which are of interest. But due to some limitations,
the insufficient diversification of buildings, the insufficient
height representation and economical aspects, the data
will not be further considered.
- Remote Sensing
The evaluation of satellite images, either in analogue or in
digital form available, is conceivable. However, the
information content and the derivable accuracy is currently
not sufficient. Next generations of earth observation
systems (Baudoin, 1995, Fritz, 1995) promise some
improvements. At present optical satellite images can be
just a cheap help for controlling the completeness of
building datasets. Satellite radar systems (interferometry)
do not seem applicable yet for this purpose.
- Laser scanner for range measurement
The evaluation of high precision elevation data will be
operational. The systems use differential GPS and an
inertial navigation system for positioning and attitude
determination. The laser scanner measures during the
aircraft flight the distances strip by strip (Lohr and Eibert,
1995). Less experience, no possibility to get additional
information, economical aspects, and missing controlling
possibilities lead to no further inspection yet.
2.3 Studies
Due to the lack of available 3D city structures and no
experiences, what kind of data would be sufficient, an order
to evaluate a testsite within the city of Munich was given.
The restrictions caused by the used fieldstrength
propagation model were: generation of raster data,
evaluation of terrain heights including all buildings as
rectangles with flat roofs and without any additional
information.
Instead of testing many several different methods it was
decided to use one approach and to manipulate this data
in order to find the appropriate parameters.
The analytical photogrammetric approach was chosen
because of its well known operational technique and
accuracy. A test site in Munich (approximately 30 km?) was
evaluated by aerial images, recorded in the scale of
1:23000 with a wide angle camera. The accuracy of the
measured terrain heights and the outlines and heights of
the buildings laid in the range of some [dm], relative height
differences within building blocks of more than one meter
had to be recorded separately. The final output raster size
was 1m by 1m, the heights were stored in [dm],
This high precision dataset was the base for further
investigations. For testing also different accuracies, the
data was manipulated. Based on the 1x1 m? data, two
further datasets have been generated by means of
resampling techniques with output pixelsizes of 5 m by 5 m
and 10 m by 10 m respectively.
In the following figure, a part of the dataset of Munich
(centre of Munich around the Frauenkirche) is displayed.
For better recognition only building heights are shown (the
darker the shading the higher the building).
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
