Figure 5: Number of primitives per netto time. Left: all prim- 
itives, right: only CSG-trees with single primitives, maximum 
number per time: 170, maximum time 250 s. 
200 
  
  
Tid ples + op rabid ples + 
130 
100 
  
  
  
  
tn aa. ^ e 
  
  
A closer look at the acquisition times reveils the following: 
e The brutto times are shorter by appr. 25 % com- 
pared to the version used a year ago within the test 
München. 
The experienced operator, however, showed a signifi- 
cant higher performance. 
e The netto times, covering the internal loop for fitting 
the primitives to the image content, however are not 
significantly different between the two operators (not 
shown) 
e [here is no significant difference in acquisition time 
between single-primitive and complex buildings (cf. 
fig 5). The median time per primitive lies in the range 
of 40 s. This time includes on-line checking using 3D- 
visualization. 75 % of all building primitives can be 
acquired in a time below a minute (cf. fig. 4) 
4 OUTLOOK 
This intermediate report on an semiautomatic system for 3D- 
data acquisition on a standard workstation with a one-eye ca- 
pability has shown the efficiency of deriving 3D-city models 
from digitized aerial images. The next step in the develop- 
ment are investigations into the usefuleness of new automa- 
tion procedures especially matching tools. Also an investi- 
gation into the accuracy of the acquired data is necessary. 
The flexibility of the setup will support further increases in 
efficiency for semiautomatic 3D-data acquisition. 
REFERENCES 
[Bignone et al., 1996] Bignone, F., Henricsson, O., Fua, P., 
and Stricker, M. (1996). Automatic Extraction of Generic 
House Roofs from High Resolution Aerial Imagery. In Com- 
puter Vision '96. 
[Braun et al., 1995] Braun, C., Kolbe, T., Lang, F., Schick- 
ler, W., Steinhage, V., Cremers, A., Forstner, W., and 
Plümer, L. (1995). Models for Photogrammetric Building 
Reconstruction. Computer & Graphics, 19(1). 
[Brunn et al., 1995] Brunn, A., Weidner, U., and Forstner, 
W. (1995). Model-based 2D-Shape Recovery. In Sagerer, 
G., P. S. and Kummert, F., editors, Mustererkennung 
1995, pages 260-268. DAGM, Springer. 
[Burns and Riseman, 1992] Burns, J. B. and Riseman, E. M. 
(1992). Matching Complex Images to Multiple 3D Objects 
using View Description Networks. In /mage Understanding 
Workshop, San Diego, California, pages 675-682. 
[Eckstein and Steger, 1996] Eckstein, W. and Steger, C. 
(1996). Fusion of Digital Terrain Models and Texture 
for Object Extraction. In 2nd Airborne Remote Sensing 
Conference, San Francisco, Proceedings. 
[Forstner and Pallaske, 1993] Forstner, W. and Pal 
laske, R. (1993). Mustererkennung und 3D- 
Geoinformationssysteme. In Proc. of 3. Int. Anwen- 
derforum für Geoinformationssysteme, Duisburg. 
[Grün et al., 1995] Grün, A., Kübler, O., and Agouris, P., 
editors (1995). Automatic Extraction of Man-Made Ob- 
Jects from Aerial and Space Images. Birkhäuser. 
[Gülch, 1996] Gülch, E. (1996). Zwischenbericht ” Interak- 
tives 3D-Erfassungssystem" . Interner Bericht, Institut für 
Photogrammetrie, Universitat Bonn. 
[Heuel and Nevatia, 1995] Heuel, S. and Nevatia, R. (1995). 
Including Interaction in an Automated Modelling System. 
In Proceedings International Symposim on Computer Vi- 
sion. IEEE. 
[Krzystek, 1991] Krzystek, P. (1991). Fully Automatic Mea- 
surement of Digital Elevation Models. In Photogrammetric 
Week, Stuttgart, Proceedings, pages 203-214. 
[Lang et al., 1995] Lang, F., Lôcherbach, T., and Schickler, 
W. (1995). A One-Eye Stereo System for Semi-Automatic 
3D-Building Extraction. Geomatics Info Magazine. 
[Lang and Schickler, 1993] Lang, F. and Schickler, W. 
(1993). Semiautomatische 2D-Gebäudeerfassung aus dig- 
italen Bildern. ZPF, 5:193-200. 
[Leberl et al., 1994] Leberl, F., Gruber, M., Uray, P., and 
Madritsch, F. (1994). Trade-Offs in the Reconstruction 
and Rendering of 3D Objects. In Kropatsch, W. and 
Bischof, H., editors, Mustererkennung, Proceedings, pages 
58-73. DAGM. 
[Liedtke et al., 1991] Liedtke, C.-E., Busch, H., and Koch, 
R. (1991). Shape Adaptation for Modelling of 3D Objects 
in Natural Scenes. In Proc. of CVPR, Hawaii. 
[Lócherbach, 1995] Lócherbach, T. (1995). System Perfor- 
mance of Semiautomatic Building Reconstruction. In 2nd 
Course on Digital Photogrammetry. Inst. f. Photogram- 
metry, Bonn. 
[Mundy et al., 1992] Mundy, J., Binford, T., Boult, T., Han- 
son, A., Beveridge, R., Haralick, R., Ramesh, V., Kohl, C., 
Lawton, D., Morgan, D., Price, K., and Strat, T. (1992). 
The Image Understanding Environment Program. In Pro- 
ceedings Computer Vision and Pattern Recognition, pages 
406-416. 
[Price and Huertas, 1992] Price, K. and Huertas, A. (1992). 
Using Perceptual Grouping to detect Objects in Aerial 
Scenes. In Internat. Archives for Photogrammerty, Comm. 
IIl, Washington, volume 29, pages 842-855. 
[Quam and Strat, 1991] Quam, L. and Strat, T. (1991). SRI 
Image Understanding Research in Cartographic Feature 
Extraction. In H. Ebner, D. Fritsch, C. H., editor, Digi- 
tal Photogrammetric Systems, pages 111-121. Wichmann, 
Karlsruhe. 
[Sester and Fórstner, 1989] Sester, M. and Forstner, W. 
(1989). Object Location Based on Uncertain Models. In 
Proc. DAGM Symposium, Hamburg. 
266 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
toi