Abstract
A novel approach to the computation of an approximate estimate of spatial object pose from camera images is proposed. The method is based on a neural network that generates pose hypotheses in real time, which can be refined by registration or tracking systems. A modification of Kohonen’s self-organizing feature map is systematically trained with computer generated object views such that it responds to a preprocessed image with one or more sets of object orientation parameters. The key concepts proposed are representations of spatial orientation that result in continuous distance measures, and the choice of a fixed network topology that is best suited to the representation of 3-D orientation. Experimental results from both simulated and real images demonstrate that a pose estimate within the accuracy requirements can be found in more than 90% of all cases. The current implementation operates at near frame rate on real world images.
now with the Signal Processing Lab of the Swiss Federal Institute of Technology in Lausanne, Switzerland.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
H. S. M. Coxeter. Regular Polytopes. Dover, 1973.
A. Khotanzad and J. Liou. Recognition and pose estimation of 3-D objects from a single 2-D perspective view by banks of neural networks. In Proc. of the Artificial Neural Networks in Engineering Conference, pages 479 – 484, 1991.
D. G. Lowe. Fitting parametrized three-dimensional models to images. IEEE Transactions on Pattern Analysis and Machine Intelligence, 13 (5): 441 – 450, May 1991.
M.-C. Lu, C.-H. Lo, and H.-S. Don. A neural network approach to 3-D object identification and pose estimation. In Int. Conf on Artificial Neural Networks, pages 2600 – 2605, 1991.
C. Maggioni and B. Wirtz. A neural net approach to 3-D pose estimation. In Proc. International Conf. on Artificial Neural Networks, pages 75 – 80, 1991.
K. Park and D. J. Cannon. Recognition and localization of a 3-D polyhedral object using a neural network. In Int. Conf. on Robotics and Automation, pages 3613 – 3618, 1996.
T. Poggio and S. Edelman. A network that learns to recognize three-dimensional objects. Nature, 343: 263 – 266, 1991.
H. J. Ritter. Learning with the self-organizing map. In Proc. International Conf. on Artificial Neural Networks, pages 379 – 384, 1991.
K. Shoemake. Animating rotations with quaternion curves. Computer Graphics,19(3):245 – 254, July 1985.
S. Winkler. Model-based pose estimation of 3-D objects from camera images using neural networks. Technical Report IB 515–96–12, Institut für Robotik und Systemdynamik. Deutsche Forschungsanstalt für Luft-und Raumfahrt, 1996. Diplomarbeit. Institut für Nachrichten-und Hochfrequenztechnik. Technische Universität Wien.
P. Wunsch and G. Hirzinger. Registration of CAD-models to images by iterative inverse perspective matching. In Int. Conf. on Pattern Recognition, pages 78 – 83, 1996.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1997 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Winkler, S., Wunsch, P., Hirzinger, G. (1997). A Feature Map Approach to Real-Time 3-D Object Pose Estimation from Single 2-D Perspective Views. In: Paulus, E., Wahl, F.M. (eds) Mustererkennung 1997. Informatik aktuell. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-60893-3_12
Download citation
DOI: https://doi.org/10.1007/978-3-642-60893-3_12
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-63426-3
Online ISBN: 978-3-642-60893-3
eBook Packages: Springer Book Archive