International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
  
resident. 
settlement 
objects 
(190) 
  
  
  
  
  
manual [classification 
OK p cx | unclear | not OK 
(172) . (18) C 2 (2) 
  
  
  
  
  
  
  
unclear 
  
  
  
  
  
Figure 10. Results 
to be captured as industrial settlement areas, but this 
characteristic cannot be seen in images) The class not OK 
contains all objects where a change in the landscape happened 
or which were captured wrongly. In this class were only 1 -2-— 
3 objects, because the GIS data was up-to-date and settlement 
objects change their object class only very seldom. 
The result of the object-based classification can also be seen in 
Figure 10. The automatic approach classified 93% (163 + 57) of 
all objects of the class OK to same object class as they were 
collected in the GIS database. The classification of the objects 
of the class unclear reflects the situation that even a human 
operator is not able to classify these objects unambiguous: 6094 
(9 * 13) objects where classified to the same object class as they 
were collected and 4095 (9 + 6) where classified to the other 
class. All objects (2 + 1) of the class not OK were classified into 
the other class, as they were collected. It is very important for a 
change detection approach that all objects, where definitely a 
change has happened, are found by the program. Otherwise an 
operator has to overwork the whole result of the automatic 
approach, which is nearly as much work as a manual change 
detection. If the automatic approach finds to many changes, is 
in opposition to that not a big problem. In our example 7% of 
the objects will be marked as a potential change by the object- 
classification and have to be controlled even there is actually no 
change. 
728 
The manual overwork of the object-based classification could 
be further decreased because the result is not only a 
classification of the objects into the most likely class but also a 
probability vector that describes the quality of the result. These 
quality measures can be used for an automatic evaluation of the 
results. This topic will be researched in future work. Another 
field of application for the automatically derived quality 
measures is the automatic quality control of GIS databases. 
5. REFERENCES 
Arbeitsgemeinschaft der Vermessungsverwaltungen der Lünder 
der Bundesrepublik Deutschland (AdV) 1988. Amtlich 
Topographisches-Kartographisches Informationssystem 
(ATKIS) Bonn. 
Benz et. al, 2004. Multi-resolution, object-oriented fuzzy 
analysis of remote sensing data for GIS-ready information. 
ISPRS Journal of Photogrammetry & Remote Sensing, 58 
(2004), 239 — 258. 
Blaschke et. al., 2000. Object-oriented image processing in an 
integrated GIS/remote sensing environment and perspectives for 
environmental applications. In: Cremers, A. and Greve, K. 
(2000) (Eds.). Environmental Information for Planning, Politics 
and the Public, Metropolis-Verlag, Marburg, Volume Il, 555 — 
579. 
Haala, N., Walter, V., 1999. Classification of urban 
environments using LIDAR and color aerial imagery. In: 
International Archives for Photogrammetry and Remote 
Sensing, Vol. XXXII, Part 7-4-3W6, 76 - 82. 
Hinz, A., Dürstel, C., Heier, H.: DMC — The Digital Sensor 
Technology of Z/I-Imaging. In: Fritsch, D. and Spiller, R.: 
Photogrammetric Week 01. Wichmann, 93 - 103. 
Schleyer, A. 2001. Das Laserscan-DGM von Baden- 
Württemberg. In: Fritsch, D. and Spiller, R.: Photogrammetric 
Week 01, Wichmann, 217 — 226. 
Suveg and Vosselman, 2002. Localisation and generation of 
building models. International Archives of Photogrammetry and 
Remote Sensing 34, Part 3A, 356 — 360. 
Walter, 1999. Comparison of the potential of different sensors 
for an automatic approach for change detection in GIS 
databases. In: Lecture Notes in Computer Science, Integrated 
Spatial Databases: Digital Images and GIS. international 
Workshop ISD '99, Springer, 47 — 63, 
Walter, 2004. Object-based classification of remote sensing data 
for change detection. ISPRS Journal of Photogrammetry & 
Remote Sensing 58 (2004), 225 - 238. 
Zhang, 2004. Towards an operational system for automated 
updating of road databases by integration. of imagery and 
geodata. ISPRS Journal of Photogrammetry & Remote Sensing 
58 (2004), 166 - 186. 
6. ACKNOWLEDGEMENTS 
The author wishes to thank the company Zl/Imaging for the 
friendly and uncomplicated provision of the test data sets. 
KEY V 
ABSTI 
The up 
urban p 
fhree t 
modelii 
method 
are app 
road ne 
features 
encomy 
been te 
imagen 
Keepin, 
many C 
traffic 1 
commu 
map, i 
image-l 
feature 
become 
spatial 
multi-sj 
cost of. 
Researc 
limited, 
extracti 
develop 
(1998). 
an exis 
the roa 
snake a 
followit 
roads, : 
(2001), 
intersec 
road da 
road seg 
the pitfz 
differen 
unchang 
deforms 
mode. 7 
detectio 
versions 
  
x Corm