International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
highly accurate 3D model is needed to ensure reliable results 
(Sauerbier and Lambers, 2003). 
1.1. Generation Systems of 3D City Models 
The needs for 3D city models are growing and expanding 
rapidly in a variety of fields. In a steady shift from traditional 
2D-GIS toward 3D-GIS, a great amount of accurate 3D city 
models have become necessary to be produced in a short period 
of time and provided widely on the market (Takase et al, 
2003).Traditional modeling method of 3D city models had 
required enormous amount of time for manual works. Ordinary 
modeling method of 3D city used to be: 
1. Scan map and get digital image, 
. Trace digital image of map with 3D CAD software resulting 
in 2D data of buildings outlines, 
3. Manually make 3D modeling of buildings with 3D CAD by 
extruding 2D outlines to building height, and/or modeling 
manually detailed 3D geometry referring to drawings and 
photographs also with 3D CAD. 
N 
Especially, manual modeling with 3D CAD software was most 
time-consuming and required operators’ expertise. Therefore 
the old method was not applicable for production of great area 
of city model in a short period of time. The system development 
of automatic generation of 3D city model aimed great saving of 
production time. The new method has realized surprising 
reduction of time in production of 3D city model through 
automatic generation programs. Moreover, efficiency in texture 
mapping as well as quality and accuracy of 3D city model has 
been greatly improved. In the 3D City Model automatic 
generation systems the material data includes laser profiler data, 
aerial image, and 2D digital map. With the material data, 
detailed and accurate 3D city model is automatically generated 
(Takase et al, 2003). Automatic reconstruction from aerial 
images, has shown promising results, however one has to note 
that often special image material has been used which is not 
available in general, for example large scale, multiple overlap 
or color images, or additional height models. Even then, the 
reliable extraction of buildings in densely build-up areas has not 
been demonstrated yet. Semi automatic approaches have been 
reported both for image -and Digital Surface Model (DSMs)- 
based systems. They can be divided into approaches which 
model buildings from a fixed set of volumetric primitives which 
are combined and approaches which build the topology of the 
surface directly (Brenner et al., 2001). 
1.2. Data Source of 3D City Models 
Various methods of data capture are available for generation of 
3D city models (Wolf, 1999); 
e Photogrammetric method: The Photogrammetric method is 
proven and provides exact and definite interpretation 
results. 
e Laser scanning method from airplanes: The laser scanner 
method provides a large quantity of unstructured elements it 
cannot be used optimally for achieving the interpretation 
processing. 
If we can extent these two titles in detail, we can category as 
aerial imagery, terrestrial imagery, satellite imagery and 
444 
laserscanner data. These are used as raw data whereas 
depending on requirements, different approaches are applied. 
1.2.1. Aerial Imagery 
Photogrammetric methods are well suited for the economic 
acquisition of 3D city models, making it possible to recover the 
structure as well as the dimensions. On the other hand, classical 
photogrammetric measurement is mostly point based, which 
does not exploit the inherent structure of buildings and thus 
cannot be optimal economically (Brenner et al., 2001). At that 
moment, aerial images are the most common used raw data. For 
capturing the 3D point cloud, the stereo pairs of the images are 
needed. The scale of the images depends on the accuracy that is 
required for the 3D model and is normally about 1:5000 with a 
forward and a side overlaps of 30 and 60 percent respectively. 
If the images are used for True-Orthophoto, the side overlap is 
suggested to be 60 percent. Using this data, many building 
details can be measured from the aerial images and the 
measurement error is maximal 0,2 meter in height (Ulm, 2003). 
1.2.2. Terrestrial Imagery 
Almost all current systems apply airborne data for the 
collection of 3D city models. Of course data capture is also 
feasible based on terrestrial images. Commercially available 
software tools allow for 3D measurement at high accuracies, 
nevertheless close range techniques for architectural 
photogrammetry currently are too time consuming for an area 
covering data collection. Airborne data is more or less 
equivalent to terrestrial images if geometric data capture is 
aspired, but the integration of terrestrial imagery is mandatory 
for applications like texture mapping (Brenner et al., 2001). 
1.2.3. Satellite imagery 
In case of large areas, recently high resolution satellite imagery 
is used, like the 1-meter panchromatic from Ikonos. The data 
capturing process is the same as with aerial images, but the 
accuracy is less, measurement error can be up to 1 meter in 
height. DTM and Orthophoto can be derived automatically 
(Ulm, 2003). 
1.2.4. Laser Scanner Data 
For the generation of 3D city models from lasers canner data, a 
density of laser scanner points of more than 2 points/sqm 
are required. Big areas are already surveyed with laser scanner, 
which is seen as an advantage for the application of this data. 
The procedure for the calculation of 3D building models from 
laser scanner data uses a tangential plane as a first 
approximation that suits the laser scanner points. From this 
geometric model, edge lines are derived whereas edge lines of 
building structures are generated (e.g. eaves lines, ridge lines 
etc.). The accuracy is expected to be 0.3-0.5 meters in height 
(Ulm, 2003). 
Automatic systems working solely on the basis of DSMs 
acquired by laser scanning have been reported. Since DSMs 
represent the geometry of the surface directly, they have 
advantages with regard to automated interpretation (Brenner et 
al., 2001). 
  
Internat 
1.3. Vis 
In the | 
visualiz: 
90). Sm 
were di: 
while te 
models 
Nowada 
shade ai 
accurate 
photogr: 
are that 
precisio 
software 
resolutic 
Level € 
visualiz: 
the higl 
After th 
visualiz 
in the fo 
e. Wir 
obje 
only 
shac 
com 
e Sha 
Cos; 
area 
arise 
e Text 
the 
in i 
orth 
one 
In genes 
helps to 
represen 
In this s 
built us 
photogr: 
first par 
two app 
The firs 
applicat 
terrestri: 
visualize 
applicat 
Univers: 
digitizec 
evaluati: 
their or 
vector « 
visualisz