t of 
ich 
7 in 
ose 
In 
tor- 
ent 
zer- 
the 
ims 
sult 
> in 
ble, 
ray 
the 
gle 
can 
it a 
ine 
ver, 
vith 
ses 
The 
and 
ion 
1 to 
ata. 
OT- 
lata 
ha 
Chis 
ap- 
VED 
M 
nsor 
olor 
ciple 
the 
1 set 
ents 
ıten- 
high 
geometrical resolution takes place. If the radiometrical 
difference between the exchange files is too large a 
radiometrical adjustment of the new data onto the 
intensity is necessary. In the next step the retrans- 
formation into the RGB-color domain is performed. 
The result is a merged set of data preserving the most 
of the spectral quality of the original data and contains 
the high resolution from SPOT data. 
27 Postprocessing 
The final processing steps of the preparation of satellite 
image maps concerns the optimal radiometrical 
contrast. For this purpose in most cases simple histo- 
gram modifications are sufficient. Furthermore special 
filters are used to enhance the edge elements in order 
to improve the visibility of details in the final product 
(TAUCH & KÄHLER 1988). 
The result of all described processing steps is a satellite 
image database which now can be used as a layer in a 
GIS or which can be completed with cartographic 
elements used as map. 
3. BASIC DATA STRUCTURE OF AN HYBRID GIS 
In order to the needs of scientists, planners and other 
GIS users for actual information, the above described 
methods have to be integrated into GIS. For this 
purpose the graphic database must include vector data 
as well as raster graphics in a so called hybrid system. 
The basic data structure of an hybrid GIS will be intro- 
duced in the following. 
The actual situation in development of GIS 
technology represents differences between the user 
requirements and the available GIS hardware and 
software. Problems occure for example in building a 
relatively large scaled hybrid GIS for a complete 
country. For the territory of the Federal Republic of 
Germany e.g. a digital vector mapping system called 
ATKIS (Common Topographic-Cartographic Infor- 
mation System) will be established. For this purpose 
nowadays it looks unlikely that a digital vector data- 
base can be build up by hand-digitizing or raster to 
vector conversion of scanned topographic maps. 
Hand-digitizing - the best way today to collect logical 
structured data - takes an enormous amount of time 
and on the other hand suitable cartographic pattern 
recognition methods are not available yet. Even for 
this task (ATKIS) an hybrid GIS represents a good 
solution, where large format scanners will be used for 
creating digital raster data as input for the GIS, and for 
map revision vector data are used as well. 
31 Data Types 
Generally the input data for hybrid systems vary in a 
large field (see figure 4). Besides scanned topographic 
maps different types of data are expected, like: satellite 
image data, digital orthophotos, image data from 
airborne scanner systems, digital elevation models, 
vector data from map digitizing, from photogramme- 
tric systems and terrestrial survey. Finally non graphic 
data complete the GIS input. These data are needed to 
support tasks like map revision, production of 
printing originals, data interpretation and analysis for 
many scientific and commercial applications, envi- 
ronmental protection e.g. 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
&H a EI 
TH al ” 0 —— 
à ais a» [si 
terrestrial photo- map alphnum. existing satellite CCD- scanner 
survey  grammetry digitizing terminal Spin camera 
ata 
vector data non graphic data raster data 
GIS 
  
  
Fig.4: GIS input data 
In context to the input data structure the GIS data 
storage works on three different data types: 
-non graphic data, 
-vector data, 
-raster data. 
Non graphic data are managed like usual in standard 
database management systems. So also during GIS 
applications query languages (SQL e.g.) are helpful 
691 
tools to solve the actual tasks. Additional connections 
between the non graphic data and graphic data are 
needed to create and represent the final result. 
Therefore the topological data structure of the vector 
data should be an object orientated model, so that 
special pointer can be used like object class, object 
number and so on. Geographic or geodetic coordinates 
are used as pointer as well, especially if raster graphics 
are needed.