facilitating
Iterest, the
easily ob-
untered in
aforemen-
(periments
. gaps, low
veness of
d road
led least
esents at
s method.
the white
/hite dots
> tracking
(position
a corner,
atching to
results
on of this
segments
raction or
failure due to its localized character.
À comparison of least squares template matching to
active contour models has also been performed and
Fig. 10 shows some examples of these experiments.
As expected, least squares template matching of-
fered a large amount of highly precise edge points
and can be very effective when the edges to be ex-
tracted are rather smooth. On the other hand, active
contour models may not be as precise but their global
nature enables them to avoid distractions due to radi-
ometric or even semantic noise.
Fig. 10: Edge detection with active contour models
(left) and least squares template matching (right)
Globally enforced least squares template matching,
presented in section 6 and currently under develop-
ment, is expected by design to fuse the advantages
of both techniques and overcome their inefficiencies.
8. CONCLUDING REMARKS
The semi-automatic object extraction strategies pre-
Sented in this paper can be ideally used for monoplot-
ting on a digital photogrammetric station. They all
have their share of merits and shortcomings which
make each of them appropriate for a more or less ex-
tensive variety of object types and/or applications. By
applying them to orthophotos tied to corresponding
DTMS, survey coordinate information can also be as-
Signed to the extracted objects. Thus, a powerful and
highly automated digital photogrammetric GIS data
capture scheme is emerging, which can substantially
improve the performance of this currently cumber-
Some and relatively costly procedure.
REFERENCES
Ballard D.H. & C.M. Brown (1982): Computer Vision,
Prentice Hall, Englewood Cliffs, NJ.
Baltsavias E.P. (1991): Multiphoto Geometrically
Constrained Matching, Mitteilungen Nr. 49, Institute
of Geodesy & Photogrammetry, ETH Zurich, Switzer-
land.
Bartels R.H., J.C. Beatty & B.A. Barsky (1987): An In-
troduction to Splines for Use in Computer Graphics
and Geometric Modelling, Morgan Kaufmann Pub-
lishers Inc., Los Altos, CA.
Fischler M.A., J.M. Tenenbaum & H.C. Wolf (1981):
Detection of Roads and Linear Structures in Low-
Resolution Aerial Imagery Using a Multisource
Knowledge Integration Technique, Computer Graph-
ics & Image Processing, Vol. 15, pp. 201-223.
Fua P. & Y.G. Leclerc (1990): Model Driven Edge De-
tection, Machine Vision & Applications, Vol. 3, pp. 45-
56.
Grossmann A. & J. Morlet (1984): Decomposition of
Hardy Functions into Square Integrable Wavelets of
Constant Shape, SIAM Journal of Mathematics, Vol.
15, pp. 723-736.
Gruen A. & E.P. Baltsavias (1988): Geometrically
Constrained Multiphoto Matching, Photogrammetric
Engineering & Remote Sensing, Vol. 54, No. 5, pp.
633-641.
Gruen A. & D. Stallmann (1992): High Accuracy Di-
mensional Measurement Using Non-Targeted Object
Features, International Archives of Photogrammetry
& Remote Sensing, Washington, DC, Vol. XXIX, Part
B5, pp. 694-700.
Gruen A. & D. Stallmann (1993): High Accuracy Edge
Matching with an Extension of the MPGC-Matching
Algorithm, Intemational Association of Geodesy
Symposia, Vol. 108, pp. 339-350.
Kass M., A. Witkin & D. Terzopoulos (1988): Snakes:
Active Contour Models, International Journal of Com-
puter Vision, Vol. 1, No. 4, pp. 321-331.
Mallat S.G. (1989): A Theory of Multiresolution Signal
Decomposition: The Wavelet Representation, IEEE
Pattern Analysis Machine & Intelligence, Vol. 11, No.
7, pp. 674-693.
Suetens P., P. Fua & A.J. Hanson (1992): Computa-
tional Strategies for Object Recognition, ACM Com-
puting Surveys, Vol. 24, No. 1, pp. 5-61.
153