photographs, using digital orthoimages and photo textures 
has some disadvantages: 
Shading and shadows within the photographs depend 
on the position of the sun at the time of exposure. 
The content of the orthoimage shows a distinct seasonal 
situation. 
Depending on the exterior orientation of the camera, 3D 
objects are displaced and occlude parts of the terrain 
surface or of other 3D objects. 
Some disadvantages when using digital orthoimages as 
texture information can be seen in Figure 2: Shadows cast 
by the wood in the lower right corner of the scene; displaced 
fruit-trees around the farm-buildings. These problems can 
not be solved entirely through 3D modeling of the objects, 
because their influence is active also in regions of the 
orthoimage near to these objects. 
Thus restoration of the orthoimage texture from shadows, 
displacements and occlusions is necessary to reach the goal 
of a photorealisitic landscape visualization (cf. section 4). 
The 3D objects can be handled similarely (cf. section 5). 
The results of these modeling steps are enhanced orthoima- 
ges and enhanced photo textures. Mapping them onto the 
geometric elements leads to textured terrain surface and 3D 
objects which build up a highly realistic scene. 
974 
  
Figure 2: Visualization based on a digital orthoimage as well as 3D modeling and photo textures for the buildings 
4. ENHANCEMENT OF THE ORTHOIMAGE 
TEXTURE 
To enhance the quality of the orthoimage texture for visua- 
lization purposes some of the situation-dependent effects in 
the orthoimage have to be replaced. All striking effects 
(shadows, displacements, occlusions) are caused by objects 
projecting from the ground. Thus, the replacement of these 
objects is the most important task. It can be subdivided into 
the recognition of 3D objects, the determination of areas 
affected by the 3D objects and the replacement of the 
affected areas (cf. Fig. 3). 
4.1 Recognition of 3D objects: 
Approximate information for the shape of 3D objects can 
be taken from the land use information, e.g. the borders of 
woods, hedges and buildings. However, for the purpose of 
land consolidation not all 3D objects are acquired. The 
fruit-trees visible in Figure 2 are for example not contained 
in the GIS data nor is their shape. Thus, methods of digital 
photogrammetry are suggested to determine the position 
and the shape of the 3D objects. 
Automatically derived DEM using image matching techni- 
ques applied to large scale stereo models contain bumps that 
represent the 3D objects. Smoothing or comparison with 
ordinary DEM (ground DEM) and fusion with orthoimage 
data improve the shape determination of the objects (Haala, 
1994, Eckstein/Steger, 1996). As automatic methods don't 
work reliable and exactly for complete scenes so far, semi- 
automatic methods (Lang/Schickler, 1993) can be used (e.g. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996