ISPRS Commission III, Vol.34, Part 3A „Photogrammetric Computer Vision“, Graz, 2002 
  
  
Figure 3: Registration ofa point cloud to a CAD model: 
Initial and final position of the point cloud 
  
T— TUUM 
T 
—©— Minimization with quadratic approximants 
-U- Minimization of point-to-point distances (ICP) d 
rer ee Pet Ee E M MESES, APT | 
25 
MS alignment error 
c m 
e 
e 
  
o 
  
  
SLT E 
S m dor rz op. 4 
6 
4 5 
Iterations 
o 
Figure 4: Comparison of the convergence rate for 
our method vs. minimization of point-to-point distances in 
each step (ICP). 
6 EXTENSIONS AND FUTURE RESEARCH 
Here we did not deal with those systematic and random 
errors in the 3D point clouds that arise in the data capturing 
process. Our algorithm for the simultaneous registration of 
multiple point clouds iteratively minimizes a function F — 
S2 UI Ok(x)o em 1,279. The weights w; can be used to 
successively downweight outliers. Appropriate weighting 
schemes may be found in (Rousseeuw Leroy, 1987). 
In our contribution we have assumed that a rough, initial 
alignment of the point clouds is given and that small dis- 
placements of the point clouds are sufficient to bring them 
in optimal alignment. A very ambitious task for future re- 
search is to derive a stable algorithm to find these rough 
initial alignments. Such an algorithm should be applicable 
to multiple point clouds and should exploit as much infor- 
mation on the involved geometric entities as possible. 
Finally, for special geometries like 3D objects composed 
of simple surfaces one may have additional or more pre- 
cise information on the squared distance function (Kverh 
Leonardis, 2002). Additional work has to be done to in- 
clude such information in the basic algorithm. 
A - 270 
ACKNOWLEDGEMENTS 
This work has been carried out within the K plus Compe- 
tence Center Advanced Computer Vision and was funded 
from the K plus program. 
REFERENCES 
Bergevin R., Laurendeau D., Poussart, D., 1996. Registering 
range views of multipart objects. Comput. Vision Image Under- 
standing, 61, pp. 540—547. 
Besl, P. J., McKay, N. D., 1992. A method for registration of 
3D shapes. /EEE Trans. Pattern Anal. and Mach. Intell., 14, pp. 
239—256. 
Blais, G., Levine, D., 1995. Registering multiview range data to 
create 3D computer objects. IEEE Trans. Pattern Anal. Mach. In- 
tell., 17, pp. 820—824. 
Bourdet, P., Clément, A., 1988. A study of optimal-criteria iden- 
tification based on the small-displacement screw model. Annals 
of the CIRP, 37, pp. 503—506. 
Chen, Y., Medioni, G., 1992. Object modelling by registration 
of multiple range images. /mage and Vision Computing, 10, pp. 
145-155. 
Dyn, N., Levin, D., Gregory, J. A., 1990. A butterfly subdivi- 
sion scheme for surface interpolation with tension control. ACM 
Transactions on Graphics, 9(2), pp. 160—169. 
Eggert, D. W., Fitzgibbon, A. W., Fisher, R. B., 1998. Simula- 
neous registration of multiple range views for use in reverse en- 
gineering of CAD models. Computer Vision and Image Under- 
standing, 69, pp. 253—272. 
Faugeras, O. D., Hebert, M., 1986. The representation, recog- 
nition, and locating of 3-D objects. Int. J. Robotic Res., 5, pp. 
27-52. 
Horn, B. K. P., 1987. Closed-form solution of absolute orientation 
using unit quaternions. J. Opt. Soc. Am. A, 4, pp. 629—642. 
Kvehr, B., Leonardis, A., 2002. A new refinement method for 
registration of range images based on segmented data. Comput- 
ing, 68(1), pp. 81-96. 
Nikolaidis, N., Pitas, L, 2001. 3-D Image Processing Algorithms. 
Wiley. 
Pottmann, H., Hofer, M., 2002. Geometry of the squared distance 
function to curves and surfaces. Proceedings Mathematics and 
Visualization, Springer, to appear. 
Pottmann, H., Wallner, J., 2001. Computational Line Geometry, 
Springer-Verlag. 
Rousseeuw, P. J., Leroy, A. M., 1987. Robust Regression and Out- 
lier Detection, Wiley. 
Rusinkiewicz, S., Levoy, M., 2001. Efficient variants of the ICP 
algorithm. In: Proc. 3rd Int. Conf. on 3D Digital Imaging and 
Modeling, Quebec. 
Yang, M., Lee, E., 1999. Segmentation of measured point data us- 
ing a parametric quadric surface approximation. C. omputer-Aided 
Design, 31, pp. 449—457. 
Zhang, Z., 1994. Iterative point matching for registration of 
smooth surfaces using differential properties. In: Proceedings of 
the 3rd European Conference on Computer Vision, Stockholm, 
pp. 397—406. 
Zorin, D., Schróder, P., Sweldens, W., 1997. Interactive multires- 
olution mesh editing. In: SIGGRAPH 97 Conference Proceed- 
ings, pp. 259—268.