can be considered for the generalization 
software of the IfK. These digital information 
can result directly from terrestrial surveying, 
photogrammetry, digitization of analogous 
maps (with digitizer or scanner plus raster- 
vector-conversion and  structuring), from 
available strange digital data or from a lot 
of other sources. 
After the selection or respectively supply 
of data first the modelling of independent 
objects with their geometry and the belonging 
descriptive attributes is necessary. In addition 
definitions as the degree of generalization, 
the output scale, the sequence of processing, 
quite a few threshold values and many more 
as well as the kind of symbolization are to be 
selected by the user. Mainly the question 
"What should be presented how in the derived 
digital cartographic model (DCM) ?" have 
to be answered. These predefinitions and 
parameters for the generalization process, 
which play a substantial role in DCM-derivation 
in all phases, can be individually specified 
by the GIS-user for different applications 
by editing a simple control file (Powitz, 
Schmidt, 1991). 
5. GENERALIZATION OF TRAFFIC NETWORK 
For a scale-dependent graphic presentation 
of GIS-objects from the feature class 
"traffic roads’ several computation steps 
are necessary in generalization processing. 
The process includes firstly the data modelling 
and after that follows the cartographic 
generalization with the symbolization. The 
modules of IfK out of the generalization 
software CHANGE for the processing of the 
traffic roads are named CHANGE. Roads. 
The first step of computer-assisted pre- 
processing includes an extensive control 
of plausibility as incorrect or illogical data 
have to be changed constructively. These 
insufficiencies can be caused by errors 
in digitization or by the inconsistency of 
different data sources etc. These data checks 
are concerning geometric as well as 
geometric-conceptual terms. For example 
unwelcome line intersections, futile loops or 
wrong attributes are tracked down and 
automatically removed. 
Following on is the design of an orderly 
topological network of the streets and ways. 
If the outlines limiting both sides of the roads 
are provided as original geometric information 
for middle axis in the sense of centre lines of 
these band shaped objects will be calculated 
automatically. Afterwards gaps and line 
crossings (over- and under-cuts) of the centre 
lines especially at road intersections and 
junctions are eliminated. The result is a 
topological network of the roads with explicit 
nodes and edges (see figure 3). 
The important simplification of the traffic road 
representations is carried out on the basis 
of the computed topological network taking 
666 
into account the predefined generalization 
parameters. Through variation and combination 
of nodes as well as through smoothing 
of edges unimportant small forms have to be 
eliminated whereas clear-cut forms are 
modified. 
The next step in processing is the raw 
symbolization. Raw symbolization means that 
corridors or areas for the road symbols are 
reserved in the presentation level. These 
corridors are formed on both sides of 
the edges or centre lines by othogonal 
widening according to the individual object 
attributes. The wideness depends on the 
space needed by each object in its full graphic 
display in the DCM. 
Finally an optimization of the raw symboli- 
zation has to ensue as in intersections and 
junctions the derived corridor’s symbol 
sectors are overlapping. To remove the 
overlapping parts a computer-assisted 
cleaning process is implemented. The result 
of the actions described is a pre-DCM which 
supplies the final graphic presentation 
wanted in combination with the graphic 
layout elements. This pre-DCM exclusively 
contains the objects of the feature class 
“traffic roads” (see figure 4). 
6. GENERALIZATION OF BUILDINGS 
Similar to the process described for the 
feature class “traffic roads” the computer- 
assisted transfer of the objects from the 
feature class “buildings” into the DCM is 
carried out in several steps by IfK's software 
modules called CHANGE. Buildings. 
As geometric information digital outlines of 
single buildings are provided (see figure 5). 
In processing of buildings the first step also 
is an extensive plausibility control. These 
difficult checks already include on one hand 
the tracking down and on the other hand the 
automatic correction of unplausible building 
outlines. After thus optimizing the digital 
GIS-data the actual steps of cartographic 
generalization start off. 
The necessary simplification of building 
outlines is an important aspect of cartographic 
generalization for large scale applications 
(Staufenbiel, 1973, Meyer, 1987). Based on 
the given thresholds for minimal dimensions 
specific to scale (thresholds of graphic 
minima sizes for single building sides and 
area parts) insignificant small forms are 
eliminated and important ones are emphasized. 
The algorithms are based on the treatment 
of small forms that can consist of separated 
building points, small sides and parts 
of building areas. Nevertheless the original 
characteristics of the buildings should 
be consistent during and of course after 
the simplification process. 
Another generalization step contains the 
combination of neighbouring buildings while 
1 EAE o i-e a § Y's 5e 
ola MA IE a." 27 1]