we can see that the feature points are computed reliable 
enough to produce a good 3D reconstruction. The autocal- 
ibrated focal length of the camera is 610.00 pixels, while 
the true focal length is listed as 685 pixels, which is within 
about 15 percent of the correct value. 
  
^u 
+ 
ay 
ge 
+ 
Figure 3: Rendering of camera positions and feature points 
10 DISCUSSION 
In this paper we have described a system which automati- 
cally computes the correspondences for an unordered set of 
overlapping images. These correspondences are then sent 
to a bundle adjustment process to compute the extrinsic 
camera parameters. This process does not require camera 
calibration, and in fact can autocalibrate the camera focal 
length. A demonstration version of this code can be found 
in http://www.cv.iit.nrc.ca/research/PVT.html. 
REFERENCES 
Baumberg, A., 2000. Wide-baseline muliple-view correspon- 
dences. In: Proc. IEEE Conference on Computer Vision and Pat- 
tern Recognition, Hilton Head, South Carolina, pp. 1774-1781. 
Bolles, R. C. and Fischler, M. A., 1981. A ransac-based approach 
to model fitting and its application to finding cylinders in range 
data. In: Seventh International Joint Conference on Artificial 
Intelligence, Vancouver, British Colombia, Canada, pp. 637—643. 
Ferrari, V., Tuytelaars, T. and Gool, L. V., 2003. Wide-baseline 
muliple-view correspondences. In: Proc. IEEE Conference on 
Computer Vision and Pattern Recognition, Madison, Wisconsin, 
pp. 348-353. 
Harris, C. and Stephens, M., 1988. A combined corner and edge 
detector. In: Proceedings of the 4th lvey Vision Conference, 
pp. 147-151. 
Hartley, R., 1995. A linear method for reconstruction from lines 
and points. In: Proceedings of the International Conference on 
Computer Vision, Cambridge, Mass., pp. 882-887. 
Hartley, R., 1997a. In defence of the 8 point algorithm. In: 
IEEE Transactions on Pattern Analysis and Machine Intelligence, 
Vol. l9number 6. 
Hartley, R., 1997b. Kruppa's equations derived from the funda- 
mental matrix. IEEE Trans. on Pattern Analysis and Machine 
Intelligence 19(2), pp. 133-135. 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part BS. Istanbul 2004 
718 
Longuet-Higgins, H., 1981. A computer algorithm for recon- 
structing a scene from two projections. Nature 293, pp. 133-135. 
Lowe, D., 1999. Object recognition from scale invariant features. 
In: International Conference on Computer Vision, IEEE Com- 
puter Society, pp. 1150—1157. 
Martinec, D. and Pajdla. T., 2002. Structure from many per- 
spective images with occlusions. In: Computer Vision - ECCV 
2002, 7th European Conference on Computer Vision, Copen- 
hagen, Denmark, May 28-31, 2002, Proceedings, Part IL, Lecture 
Notes in Computer Science, Vol. 2351, Springer, pp. 355-369. 
Mendonca, P. and Cipolla, R., 1999. A simple technique for self- 
calibration. In: Proceedings of IEEE Conference on Computer 
Vision and Pattern Recognition, Fort Collins, Colorado, pp. 112- 
116. 
Mikolajczk, K. and Schmid, C., 2003. A performance evaluation 
of local descriptors. In: Proc. IEEE Conference on Computer 
Vision and Pattern Recognition. 
Photomodeler by EOS 
http:/www,photomodeler.com. 
Systems Inc., n.d. 
Pritchett, P. and Zisserman, A., 1998. Wide baseline stereo 
matching. In: Proc. International Conference on Computer Vi- 
sion, Bombay, India. 
Roth, G., 2002. Some improvements on two autocalibratio algo- 
rithms based on the fundamental matrix. In: International Con- 
ference on Pattern Recognition: Pattern Recognition Systems and 
Applications, Quebec City, Canada. 
Roth, G. and Levine, M. D., 1993. Extracting geometric primi- 
tives. Computer Vision, Graphics and Image Processing: Image 
Understanding 58(1), pp. 1—22. 
Roth, G. and Whitehead, A., 2000. Using projective vision to 
find camera positions in an image seqeuence. In: Vision Interface 
2000, Montreal, Canada, pp. 87-94. 
Rousseeuw, P. J., 1984. Least median of squares regression. Jour- 
nal of American Statistical Association 79(388), pp. 871—880. 
Schaffalitzky, F. and Zisserman, A., 2002. Multi-view matching 
for unordered image sets. In: Proceedings of the 7th European 
Conference on Computer Vision, Copenhagen,Denmark. 
Shashua, A., 1995. Algebraic functions for recognition. IEEE 
Transactions on Pattern Analysis and Machine Intelligence 17(8), 
pp. 779—789. 
Spetsakis, M. and Aloimonos, J., 1990. Structure from motion 
using line correspondences. International Journal of Computer 
Vision 4(3), pp. 171-183. 
Torr, P. and Murray, D., 1993. Outlier detection and motion seg- 
mentation. In: Sensor Fusion VI, Vol. 2059, pp. 432-443. 
Xu, G. and Zhang, Z., 1996. Epipolar geometry in stereo, motion 
and object recognition. Kluwer Academic.