xels are
> + F'N)
FN)
Yypothe-
ormance
'al struc-
nents of
‚ely high
ny build-
ver right
red prim-
d peaked
building
intensity
cet.
hypothe-
the per-
; at least
the high
flect this
rh quality
amounts
e entirely
formance
chniques,
Figure 8: PIVOT results on RADT5WOB
this fully automated monocular system is able to achieve ro-
bust performance on images taken from widely differing view-
points, with large variations in image photometry and scene
content.
7 CONCLUSIONS AND FUTURE WORK
Future work on PIVOT is directed along several research av-
enues, many of which have been mentioned in this paper: a
generalized attachment strategy for primitives to handle par-
tial alignments of primitives; extrusion of primitives along all
vanishing lines, not just in the vertical direction; the addition
of trihedral vertices as a supplementary intermediate feature:
the ability to handle severe low-level edge fragmentation; and
capabilities for interfacing with semi-automated systems.
While PIVOT is still under development, current results il-
lustrate the power and potential of the thorough integration
of photogrammetric methods in automated feature extrac-
tion algorithms for aerial image analysis. The combination of
computer vision techniques with a rigorous central projection
camera model leads to superior building extraction perfor-
mance on a wide variety of scenes, a necessity for the use
79
evaluation building branch | miss | quality
detection % | factor | factor %
2D 90.7 0.24 0.10 74.3
3D 84.4 0.30 0.18 67.1
Table 9: Evaluation results for RADT5WOB
of automated systems for the timely construction of highly
detailed spatial databases.
8 ACKNOWLEDGEMENTS
| wish to thank the members of the Digital Mapping Labora-
tory, particularly Steve Cochran, Yuan Hsieh, Chris McGlone,
Dave McKeown, and Michel Roux, for many insightful dis-
cussions on issues in automated building extraction.
REFERENCES
[Barnard, 1983] S. Barnard. Interpreting perspective images.
Artificial Intelligence, 21:435—462, 1983.
[Biederman, 1985] I. Biederman. Human image understand-
ing: Recent research and a theory. Computer Vision,
Graphics, and Image Processing, 32:29-73, 1985.
[Braun et al., 1995] C. Braun, T. H. Kolbe, F. Lang,
W. Schickler, V. Steinhage, A. B. Cremers, W. Forstner,
and L. Plümer. Models for photogrammetric building re-
construction. Computers and Graphics (UK), 19(1):109-
118, 1995.
[Fórstner, 1995] W. Fórstner. Mid-level vision processes for
automatic building extraction. In Proceedings: Ascona
Workshop on Automatic Extraction of Man-Made Objects
from Aerial and Space Images, Ascona, Switzerland, 1995.
[Fritsch et a/., 1994] D. Fritsch, M. Sester, and T. Schenk.
Test on image understanding. |n Proceedings: ISPRS
Commission HI Symposium on Spatial Information from
Digital Photogrammetry and Computer Vision, Volume 30,
Part 3/1, pages 243-248, Munich, Germany, 1994.
[Hsieh, 1996] Y. Hsieh. Design and evaluation of a semi-
automated site modeling system. In Proceedings of the
ARPA Image Understanding Workshop, Palm Springs, Cal-
ifornia, February 1996.
[Irvin and McKeown, 1989] R. B. Irvin and D. M. McKe-
own. Methods for exploiting the relationship between
buildings and their shadows in aerial imagery. IEEE Trans-
actions on Systems, Man, and Cybernetics, 19(6):1564-
1575, November 1989.
[McGlone and Shufelt, 1994] J. C. McGlone and J. A.
Shufelt. Projective and object space geometry for monoc-
ular building extraction. In Proceedings: IEEE Conference
on Computer Vision and Pattern Recognition, pages 54—
61, June 1994.
[McKeown and McGlone, 1993] D. M. McKeown and J. C.
McGlone. Integration of photogrammetric cues into car-
tographic feature extraction. In Proceedings: SPIE Con-
ference on Integrating Photogrammetric Techniques with
Scene Analysis and Machine Vision, Volume 1944, pages
2-15, April 14-15 1993.
[Shufelt, 1996] J. A. Shufelt. Performance evaluation and
analysis of vanishing point detection techniques. In Pro-
ceedings of the ARPA Image Understanding Workshop,
Palm Springs, California, February 1996.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B6. Vienna 1996
