Andreas Keim 
  
labelled, generalized employing the Douglas-Peuker algorithm and finally converted to contour-lines in DXF format. 
The 3-dimensional nature of the terrain can be visualized by simple hill-shading, which was rendered automatically 
from the same DEM. 
Visual SAR image interpretation needs lots of experience and knowledge about the SAR imaging principles and 
geometry, because they obey other regularities than optical sensors. Before each interpretation an interpretation key has 
to be elaborated and controlled. In this key all the derivable objects in the data are described. Also in times where 
analysis of remote sensing images by digital methods increases, the visual information extraction using human 
interpretation ability is still significant. Till this day the visual image interpretation is the most accurate method of 
information extraction. On the other hand the method is very cost expensive and time-consuming. For this investigation 
classification was mainly applied for forests, open areas and large rivers and lakes. Objects like build-up areas, road 
networks, single buildings and other cartographic signatures had to be interpreted and digitized by the operator. In some 
cases it is difficult to distinguish small rivers from roads. Therefore the information provided by contour-lines are very 
helpful and instructive. After extracting the interpreted map objects by screen digitizing in a simplified vector graphic 
(Fig: 5.2), the layout graphic is transformed into the final map graphic (Fig. 5.3). 
All the additional information, not extractable from remote sensing data but important for understanding maps, like map 
scripting must be taken from other sources e.g. Internet, other maps, or on site. Figure 4.1 to 4.3 show the steps from the 
SAR data to be interpreted, the three classified classes and the final map completed with visually interpreted objects. 
The generated data are combined in a professional WYSIWYG graphic software and prepared for printing. 
5 MAP SAMPLES 
After processing the InSAR data, described in chapter 3, the following map extracts were generated by visual image 
interpretation (Fig. 5) or by a combination of classification and visual interpretation (Fig. 4). 
  
Figure 4.1: Geocoded SAR ortho image, 
4.0 km by 2.5 km, scale 1: 50,000 
  
PUERTO = 
UCHO 
   
Figure 4.2: Classification result: Forest...green Figure 4.3: Topographic map, scale 1 : 50,000 
(medium grey), open area...yellow (light grey) 
and water...blue (upper left corner) 
  
176 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part Bl. Amsterdam 2000. 
  
  
Int 
pre: 
avai 
Sys 
in fi 
199 
sum 
pros 
RE 
Hut 
date 
Hof 
Kei 
Bilc 
Lab