ed 
his 
se- 
are 
ro- 
(th 
ole, 
on, 
ya 
  
  
ha- 
raffic 
s are 
  
* Data preprocessing 
* Center-axis determination 
* Creation of a topological network 
* Simplification of this network using different methods 
(spline or polynomial approximation, Douglas-Paucker 
filtering, highpass filter) 
* Derivation of enlarged double-line roads 
* Clarification of the interchange areas 
The user can combine these submodules in any meaningful way. 
The following input data types are legal: 
* Areas 
e Lines 
* Outline polygons (left and right road edges geometri- 
cally separated but semantically linked in one object). 
The same data types can also be put out with any combination of 
input and output data types being possible. 
The potential of this module vastly exceeds pure road generaliza- 
tion. This is shown by the following list of possible applications: 
* Generalization of double-line traffic routes to enlarged 
double-line or single-line traffic routes 
* Simplification of line data (e.g. contour lines) without 
the third dimension being lost 
* Road axis determination 
* Determination of the network topology of an urban 
road network to answer questions like "Which is the 
best way from A to B?". 
  
  
  
  
  
2 
Na Ay SN 
Fig. 5a Base map 1:5000 compiled by photogrammetric data 
acquisition with P3 Planicomp and PHOCUS 
  
  
  
  
  
Fig 5b Result of automatic generalization and interactive 
revision for a map 1:20 000 
  
  
  
Fig. 5 Example 2 
141 
  
  
  
  
  
  
  
  
CONTOUR LINES Original Data, Selection, Smoothing, 
Line Interruption, Annotation 
[ mmm | 
EN 
deu 
ee 
TRAFFIC ROUTES Original Data, Type Grouping, Deter- 
mination of Axis, Network Topology 
1 
I 
| i 
| / 
\ / 
X / 
= A 
~ 
v. 
§ FF 
DISPLACEMENT Original Data, Generalization of Traffic 
Routes and Buildings, Automatic Iden- 
tification, Interactive Displacement 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
- 
  
  
  
  
  
  
  
  
  
  
  
Fig. 3 Generalization of contour lines and traffic routes 
3.4 Sample results 
Figures 4 and 5 illustrate the potential of the PHOCUS generalization 
package in a striking way. A complex practical example contains 
densely built-up urban areas and sparsely built-up marginal areas. 
The original data set consists of about 800 objects with about 10.000 
coordinates. The derived result data set still contains about 300 
objects and about 4.000 coordinates. These numbers only refer to 
the object types 'building' and ‘road’. Generalization of contour lines 
and water bodies provides about the same results. 
4. PROSPECTS 
4.1 Future developments 
The next large complex that will be dealt with within the scope of 
CHANGE is displacement automation. This is a complicated pro- 
cedure that must allow for the interaction between individual objects. 
Once this problem has been solved, the degree of generalization 
automation attainable with PHOCUS will progress from the current 
80 % to 95 % and more. 
The system is open for additional modules that can be integrated 
with little effort. 
4.2 Conceptual generalization 
The major currently available concept generalization options are: 
* Suppression of the combination of objects belonging 
to different classes resp. object and object item types 
* Suppression of the elimination of important objects 
which. would have to be eliminated by geometrical reasons 
* Use of the attributes of the largest object for the combination