International Archives of the Photogrammetry, Remote Sensing
e Guidelines for a matching algorithm. For some appli-
cations, we can suppose that the reliability of the viewing
parameters and the 3D reconstruction is fixed. Hence, our
bundle adjustment algorithm with error propagation can as-
sess when the reliability goal is reached. If not, the match-
ing algorithm still has to extract and match more features in
order to increase the reliability in the given parameters.
e Dealing with large scale surveys. The experiments were
performed with a reduced amount of data (20-30 images,
several hundreds of measures). Dealing with large-scale
surveys will necessitate an optimized design and we will
have to handle sparse matrix. Some optimization processes
are well known in the photogrammetric community and de-
scribed in (Triggs et al., 2000).
e Spatial reasoning. A very promising area of research con-
cerns the way to choose points and lines in images in order
to improve the accuracy of the estimated parameters. As a
matter of fact, the spatial distribution of the features (points
and lines) inside the images, as well as the geometric ar-
rangement of the camera, have a tremendous impact on the
accuracy of the estimation of the different parameters of a
bundle adjustment problem (Michaelsen and Stilla, 2003).
Correlation analysis between different parameters will be of
interest in order to estimate the crucial parameters.
Introducing more knowledge concerning the data. More
constraints or parameters can be introduced depending on
the knowledge we have of the scene. For instance, vertical
lines can be introduced in the bundle adjustment thanks to
vanishing points or constraints on the 3D lines (Bentrah et
al.. 2004), (Van den Heuvel, 2001).
REFERENCES
Bentrah, O.. Paparoditis, N. and Pierrot-Deseilligny, M., 2004.
Stereopolis: An image based urban environments modeling sys-
tem. In: MMT 2004. The 4th International Symposium on Mo-
bile Mapping Technology, Kunming, China.
Chou, J. and Kamel, M., 1991. Finding the position and orien-
tation of a sensor on a robot manipulator using quaternions. The
International Journal of Robotics Research 10(3), pp. 240—254.
Faugeras, O., Luong, Q. and Papadopoulo, T., 2001. The Geom-
etry of Multiple Images. The MIT Press.
Fórstner, W., 2004. Uncertainty and projective geometry. In:
E. Bayro-Corrochano (ed.), Handbook of Computational Geom-
etry for Pattern Recognition, Computer Vision, Neurocomputing
and Robotics, Springer.
Habib, A., 1999. Aerial triangulation using point and linear fea-
tures. In: ISPRSGIS99, pp. 137-142.
Hartley, R. and Zisserman, A., 2002. Multiple View Geometry in
Computer Vision. Cambridge Univ. Press. HAR 12 02:1 L.Ex.
Heuel. S.. 2001. Points, lines and planes and their optimal estima-
tion. In: Pattern Recognition, 23rd DAGM Symposium, LNCS,
Springer, pp. 92-99.
Kraus, K., 1993. Photogrammetry. Dmmler/Bonn.
Kumar, R. and Hanson, A., 1994. Robust methods for estimating
pose and a sensitivity analysis. CVGIP 60(3), pp. 313-342.
Michaelsen, E. and Stilla, U., 2003. Good sample consensus esti-
mation of 2d-homographies for vehicle movement detection from
thermal videos. In: R. S. International Archieves of the Pho-
togrammetry and S. I. Sciences (eds), Proceedings of the ISPRS
Conference on Photogrammetric Image Analysis PIA'03, Mu-
nich, Vol. 34, Part 3/W8, pp. 125-130.
Taillandier, F. and Deriche, R., 2002. Reconstruction of 3D linear
primitives from multiple views for urban areas modelisation. In:
PCV'02, Vol. 34:3B, pp. 267-272.
and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
Torr, P. H. S. and Davidson, C., 2003. Impsac: A synthesis of
importance sampling and random sample census. PAMI 25(3),
pp. 354-365.
Triggs. B., McLauchlan, P., Hartley, R. and Fitzgibbon, A., 2000.
Bundle adjustment — A modern synthesis. In: W. Triggs, A. Zis-
serman and R. Szeliski (eds), Vision Algorithms: Theory and
Practice, LNCS, Springer Verlag, pp. 298-375.
Van den Heuvel. F. A., 2001. Towards automatic relative orien-
tation for architectural photogrammetry. In: ISPRS Commission
V, WG V/2.
Xu, G. and Zhang, Z., 1996. Epipolar Geometry in Stereo, Mo-
tion and Object Recognition. Kluwer Academic Publisher.
* Institut
^ INRIA Rhóne-
KEY WORDS:
Short Stereoscopi
ABSTRACT:
In this paper we ir
OPOLIS develop
erencing device. \
images of the ster
as a photogramme
acquisitions (t) an
clouds. We achiev
of tie points and se
and the relative po
1 USING IMAG
Many works of th
subject of the prod
tem mounted on a
vironments. The s
mate the localisatic
with direct navigat
image poses provic
tered due to GPS 1
age as a help for lc
community which
photogrammetric a
cessin is not requir
device.
In this paper we in
photographies acqi
describe an image |
relative and partial
multi-view rig.
2
The Stereopolis is t
oratory of IGN in t
of georeferenced te:
lis system as seen ii
three pairs of 4k + .
(2 GPS with choke
fectly synchronous
of georeferencing a
imaging geometry €
The two frontal fa
baseline (1.5 meter
tures (lampposts, tr
) and two short ste
of the vehicle to su
The short vertical st
augment the field o
mm focal length len
at a distance of 6 nx
In the scope of the
this system to estim: