11S
sic
5),
1S.
by
ng
ti-
ed
ng
strategy together with basic results are given. The enhance-
ment of the orthoimage used as texture and the photo-
realistic visualization of 3D objects are presented in sections
4 and 5. Concluding remarks and an outlook are given in
section 6.
2. DATA FOR SCENE DESCRIPTION
Photorealistic visualization requires high effort for geo-
metric and radiometric modeling of the scene, i.e. it strongly
depends on the available data representing the landscape as
good as possible. Using data bases for DEM and land use
based on GIS is a proper starting point. However, 3D
models for objects projecting from the terrain surface like
buildings or trees are not yet available in existing DEM and
GIS systems, although there are some considerations to
achieve this goal (Fritsch/Schmidt, 1994).
Data acquisition must be done seperately for the terrain
surface and the 3D objects. The best possible data source is
often photogrammetry, because from aerial and terrestrial
photographs both geometry and texture of the objects can
be taken. Automatic data capturing with methods of digital
photogrammetry like image matching and image under-
standing is increasingly practicable for this purpose
(Grau/Tónjes, 1994, Collins et al, 1995, Lin et al, 1995).
In the task of planning for land consolidation digital data
are used more and more (Stark/Eder, 1992). DEM, stereo-
models from aerial photographs of large scale, land use
information and digital orthoimages are available. For this
paper existing digital data (DEM and GIS) as well as color
aerial photographs of scale 1:4000 were used. These data
were provided for a land consolidation project around the
village Marchertsreuth in Bavaria. By scanning the aerial
photographs and calculating digital orthoimages rectified
textures were generated.
For 3D objects whose vertical sides were as usual not visible
in the aerial photographs, it was necessary to take additional
terrestrial photographs. In rural areas these are in principal
farm buildings, woods, hedges and trees. As these terrestrial
images are taken for the derivation of digital texture infor-
mation, the use of a CCD-Camera is suggested. For this
paper an analog camera was used. Digital information was
received by scanning the photographs (photo-CD).
3. MODELING STRATEGY
Sophisticated scene modeling is the most important task to
reach photorealistic visualizations. The main components
of the modeling strategy proposed in this paper are shown
in Figure 1 (the input data are shown in the first row).
The 3D scene can be geometrically subdivided into the
terrain surface as described by the DEM and the 3D objects
973
projecting from the terrain. In 3D computer graphics objects
like the terrain surface usually are described as triangular
strips or as quadrilateral meshes. In general the data structu-
re of a DEM is very similar and can easily be transformed
to this description. However, tiny structures are not con-
tained in the DEM information. A geometric description of
these structures requires a huge effort of data capturing and
modeling.
As photographs of the terrain represent besides radiometric
information implicitly many geometric terrain features,
mapping texture information based on digital images is a
good idea (Eder et al, 1993, Leberl et al, 1994). It clearly
increases the degree of realism for landscape visualization.
Mapping texture information to the geometry of 3D objects
aims at the same advantages (see also section 5). Therefore
the basic modeling components, shown in the second row
of Figure 1, are the terrain surface combined with a digital
orthoimage and the 3D objects combined with the photo
textures attached to their vertical walls.
The view in Figure 2 is a first result based on this informa-
tion. Similar visualizations have already been introduced by
various authors (Grau/Tónjes, 1994, Faust et al, 1994).
Tools for mapping of digital orthoimages or satellite images
onto the DEM are also integrated into various GIS.
As terrain visualisation by methods of computer graphics
should be independent of the orientation and date of the
Terr. Aerial
photogr. photogr.
Photo 3D : Digital
texture — objects orthoimage .
Enhanced
| photo
texture
Figure 1: Components of modeling strategy
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
