5. CONCLUSIONS
We have discussed the issues involved in choosing matching
techniques for different tasks and illustrated with some
selected examples. Matching of images with other images or
with maps and models in complex environments remains a
difficult and challenging task. It is this author’s view that as
the scene complexity increases, the matching problem can
not be adequately solved without using more context and
computing higher level descriptions from images, perhaps
upto the point of finding the 3-D objects themselves or larger
parts of their surfaces. Finding such objects, of course, is a
difficult problem in itself and requires major advances in the
techniques for perceptual grouping and scene segmentation.
Fortunately, it appears that, in many cases, the segmentation
and grouping processes can cooperate with the matching
processes, reducing the complexity of both.
6. ACKNOWLEDGEMENTS
I would like to thank Prof. Heinrich Ebner for inviting me to
present this paper. Andres Huertas and Sanjay Noronha have
helped me in preparation of this paper and provided the
examples used herein. The research reported here was
supported by the Image Understanding program of the
Defense Advanced Research Projects Agency (DARPA)
under grant number F49620-95-1-0457 and contract number
DACA76-93-C-0014 monitored by the Air Force Office of
Scientific Research and by the Topographic Engineering
Center respectively.
REFERENCES
[1] Ayache, N. J., and Lustman, E, 1991, Trinocular
stereo vision for robotics", Transactions of IEEE on
Pattern Analysis and Machine Intelligence, 13(1),
pp. 73-85.
[2] Bejanin, M., Huertas, A., Medioni, G., and Neva-
tia, R., 1994, Model validation for change detec-
tion", Proceedings of Worshop on Applications of
Computer Vision, Sarasota Florida, pp. 160-167.
[3] Boyer, K. L., and Kak, A. C., 1988, Structural ste-
reopsis for 3-d vision, IEEE Transactions on Pattern
Analysis and Machine Intelligence, 10, pp. 144-
166.
[4] Chung, R. C. K. and Nevatia, R., 1992, Recovering
building structures from stereo, IEEE Proceedings
of Workshop on Applications of Computer Vision,
pp. 64-73.
[5] Dhond, U.R., and Aggarwal, J. K., 1989, Structure
from stereo - a review, IEEE Transactions on Sys-
tems, Man and Cybernetics, 19(5), pp. 1489-1510.
[6] Ebner, H., Heipke, C. and Holm, M., 1993, Global
image matching and surface reconstruction in ob-
ject space using aerial images, in Barrett E. B.,
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
McKeown D. M. (Eds.), Integrating photogram-
metric techniques with scene analysis and machine
vision, Proceedings of SPIE, pp.44-57.
[7] Forstner, W., 1982, On the geometric precision of
digital correlation, International Archives for Pho-
togrammetry and Remote Sensing, 24(3), pp.176-
189.
[8] Grewe, L. L. and Kak, A. C. 1994, Stereo vision,
in Chien, Y. T. (Eds.) Handbook of Pattern Recog-
nition and Image Processing: Computer Vision, pp.
239-317.
[9] Hannah, M. J. 1989, A system for digital stereo
matching, PE & RS, 55(2), pp. 1765-1770.
[10] Heipke, C. 1996, Overview of image matching
techniques, OEFPE workshop on applications of
digital photogrammetric workstations, Lausanne.
[11] Huttenlocher, D. P. and Ullman, S., 1990, Recog-
nizing solid objects by alignment with an image, In-
ternational Journal of Computer Vision, 5(2), pp.
195-212.
[12] Lim, H. S. and Binford, T. O., 1987, Stereo corre-
spondence: a hierarchical approach, Proceedings of
Image Understanding Workshop, pp 234-241.
[13] Lowe, D. G., 1987, Three-dimensional object rec-
ognition from single two-dimensional images", Ar-
tificial Intelligence, 31(3), pp. 355-395.
[14] Medioni, G. and Huertas, A., 1991, Automatic reg-
istration of color separation films, Machine Vision
and Applications 4, pp. 33-51.
[15] Medioni, G. and Nevatia, R., 1985, Segment-based
stereo matching, Computer Vision Graphics and
Image Processing, 31(1), pp. 2-18.
[16] Mohan R. and Nevatia R., "Using perceptual orga-
nization to extract 3-D structures", IEEE Transac-
tions on Pattern Analysis and Machine Intelligence,
11(11), 1121-1139, Nov 1989.
[17] Noronha, S. and Nevatia, R., 1996, Detection and
description of buildings from multiple aerial imag-
es, Proceedings of Image Understanding Work-
shop, Palm Springs CA, pp. 469-478.
[18] Roux, M. and McKeown, D. M., 1994, Feature
matching for building extraction from multiple
views, IEEE Proceedings of Computer Vision and
Pattern Recognition, pp. 46-53.
574
KEY !
ABST
If suffi
matrix
matrix
algorit
recove
spanne
that is
the op
focal |
Relativ
ters de
two C:
image
metris!
commu
relative
tation
algorit
and re
constr:
puting
pair, e
from e
finding
of the
availab
matrix
line (se
this).
Given
calibra
Here, t
and rec
ie to
extrins
this pre
a mode
pinhole
a genet
trix [10
Is a fui
Even if
stereo
ously a
mental
penden
same s
