2004 International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
  
  
  
ts is In the feature type domain of transportation, the ATKIS-object 
catalogue of the base-DLM holds a feature type group called 
road traffic, which describes feature types like roads, paths and 
the lanes basically in terms of linear objects. Attributes like 
jects function, width or number of lanes express further 
of the characteristics. 
r the 
sides 3.12 Digital Terrain Models: In comparison to classic 
; the stereo Photogrammetry laser scanning is an efficient alternative 
ng a for generating three dimensional surface information 
ment (Knabenschuh, 1999). For a detailed description of this method 
  
refer to Baltasavias (1999). Different products are derived from 
the observed raw data. One of these is the Digital Surface 
  
tis a Model (DSM) which includes points from the ground surface as 
- the well as points from objects on top of the surface like vegetation 
quare or buildings. After classification and selection of the ground 
te on points, a high quality Digital Terrain Model (DTM) is derived. 
000). The point density of this data sets is up to four points per square 
bject meter. The planimetric accuracy is circa 0.5 m (Baltasavias, 
7). 1999b; Lohr, 1999), whereas the accuracy in height is stated 
with 0.15 m (Briese et al., 2001; Wever & Lindenberger, 1999). 
ats-is The data set which was used for the experiments represents a 
on of part of the city of Stuttgart, with a resolution of 1 m. 
nann, 
relief 32 Enhancement of street objects and arising problems 
| they 
ts are Based on the street objects given in the two dimensional vector 
data set, the streets are broadened to both sides of their middle 
axes. The width of the broadened streets is arbitrary. For each 
  
  
raster raster cell in the buffer created around the axes of the streets, 
ween the height value is recalculated based on the course of the street 
jects, axes. In figure 1 the original dataset is compared to the results 
. For of the enhancement for a section from the DTM of the city of 
yoical Stuttgart. The top image shows the scene with 1000 000 
t (Li, triangles. In the middle, the number of triangles is reduced to Figure 1: Section from the DTM of Stuttgart; top: 1 000 000 
ted in 10 000. It becomes apparent, that in the simplified mesh, the triangles, middle: 10 000 triangles, bottom: 10 000 triangles, 
h for streets are not entirely cognizable. Especially in the centre of ~~ broadened streets. 
the image, the course of the street winding up the hill is not 
perceptible. The image at the bottom demonstrates that the 
perceivableness of the enhanced street is superior to the 
representation without enhancement. 
DTM Altering the height values within the broadeneli streets without 
sized. considering the adjacent cell values brings about discontinuities 
in the between the modified cells and the bordering areas. This is 
t they visible in the top image of figure 2, where a closer view of the 
nce to broadened streets is shown. The margins of the widened streets 
appear like perforated with toothed edges. 
The first idea for solving this problem led to the simple solution 
of adjusting thc regions next to the widened streets using 
  
raphic interpolation methods within a given buffer zone. This results in 
ederal an optically smoother transition from the emphasized street to 
0 and the surrounding areas. For the adjustment, different 
yducts interpolation methods can be used. Possible is linear 
Model interpolation as well as different sigmoid weighting functions 
ogical between the border of the widened street and the original cell 
erties value in a given distance (Hatger & Kremeike, 2003). In this 
je are approach, the original values in the region of adjustment next to 
ptions the broadened street are not taken into consideration, they 
e held simply disappear from the data set. The result of the linear 
tively interpolation is shown in the bottom image of figure 2. Here the Figure 2: Section from the DTM Stuttgart; resolution 1 m; 
taken streets are widened by 10 m to both sides and the adjustment top: streets broadened around middle axes, bottom: bordering 
metric took place within another 10m. This relatively strong regions adjusted with linear interpolation within a given buffer 
enhancement was chosen to ensure the cognition for an zone. 
overview illustration. 
405