Abstract
A scheme for recognition of 3D objects from single 2D images is introduced. An object is modeled in this scheme by a small set of its views with the correspondence between the views. Novel views of the object are obtained by linearly combining the model views. The scheme accurately handles rigid objects under weak-perspective projection, and it is extended to handle rigid objects with smooth bounding surfaces and articulated objects. Unlike in other schemes, explicit 3D representations of the objects are not used. The presented scheme can be used both under an alignment framework and as a means for deriving object-specific invariant functions for indexing. Under an alignment framework, given a model and an image, the coefficients of the linear combination that aligns the model with the image need to be recovered. A small number of points in the image and their corresponding points in the model can be used for this purpose, or a search can be conducted in the space of possible coefficients. Alternatively, the scheme can be used to derive functions that are invariant to viewpoint changes of a specific object. A number of such functions are derived in this paper.
Preview
Unable to display preview. Download preview PDF.
References
Y. S. Abu-Mostafa and D. Pslatis: Optical neural computing, Scientific American 256 (1987) 66–73.
R. Basri and S. Ullman: The alignment of objects with smooth surfaces. Computer Vision, Graphics, and Image Processing: Image Understanding, 57(3), (1993) pp 331–345.
Basri, R.: Viewer-centered representations in object recognition: a computational approach. In C. H. Chen, L. F. Pau and P. S. P. Wang (Eds.), Handbook of Pattern Recognition and Computer Vision, World Scientific Publishing Company, Singapore (to appear).
Burns, J., Weiss, R., and Riseman, E.: The non-existence of general-case view-invariants, Geometric Invariance in Computer Vision, edited by J. Mundy and A. Zisserman, MIT Press, Cambridge (1992).
Chien, C. H. and Aggarwal, J. K.: Shape recognition from single silhouette. Proc. of 1st Int. Conf. on Computer Vision, London (1987) 481–490.
Clemens, D. and Jacobs, D.: Space and time bounds on model indexing, IEEE Transactions on Pattern Analysis and Machine Intelligence, 13(10) (1991) 1007–1018.
Faugeras, O. D. and Hebert, M.: The representation, recognition and location of 3D objects. Int. J. Robotics Research 5(3) (1986) 27–52.
Fischler, M. A. and Bolles, R. C.: Random sample consensus: a paradigm for model fitting with application to image analysis and automated cartography. Com. of the A.C.M. 24(6) (1981) 381–395.
Forsyth, D., Mundy, J. L., Zisserman, A., Coelho, C., Heller, A., and Rothwell, C.: Invariant descriptors for 3-D object recognition and pose. IEEE Transactions on Pattern Analysis and Machine Intelligence, 13 (1991) 971–991.
Huang, T. S. and Lee, C. H.: Motion and structure from orthographic projections. IEEE Trans. on Pattern Analysis and Machine Intelligence 11(5) (1989) 536–540.
Huttenlocher, D. P., and Ullman, S.: Recognizing solid objects by alignment with an image, Int. J. Computer Vision 5(2) (1990) 195–212.
Jacobs, D.: Space efficient 3D model indexing, IEEE Conference on Computer Vision and Pattern Recognition, (1992) 439–444.
Koenderink, J. and van Doorn, A.: Affine structure from motion, Journal of the Optical Society of America, 8(2) (1991) 377–385.
Lamdan, Y., Schwartz, J. T., and Wolfson, H.: On recognition of 3-D objects from 2-D images. Courant Inst. of Math. Sci., Rob. TR 122 (1987).
Lowe, D. G.: Three-dimensional object recognition from single two-dimensional images. Courant Inst. of Math. Sci., Rob. TR 202 (1985).
Moses, Y. and Ullman, S.: Limitations of non model-based recognition schemes, Second European Conference on Computer Vision (1992) 820–828.
Poggio, T.: 3D object recognition: on a result by Basri and Ullman, TR 9005-03, IRST, Povo, Italy (1990).
Poggio, T. and Edelman, S.: A network that learns to recognize three-dimensional objects, Nature 343 (1990) 263–266.
Poggio, T. and Girosi, F.: Regularization algorithms for learning that are equivalent to multilayer networks, Science 247 (1990) 978–982.
Rothwell, C. A., Forsyth, D. A., Zisserman, A., Mundy, J. L.: Extracting projective structure from single perspective view of 3D point sets. Proc. of 4th Int. Conf. on Computer Vision, Berlin, Germany (1993) 573–582.
Thompson, D. W. and Mundy, J. L.: Three dimensional model matching from an unconstrained viewpoint. Proc. of IEEE Int. Conf. on robotics and Automation (1987) 208–220.
Tomasi, C. and Kanade, T.: Factoring image sequences into shape and motion, IEEE Workshop on Visual motion, Princeton, NJ (1991) 21–29.
Ullman, S.: The interpretation of visual motion. M.I.T. Press, Cambridge, MA (1979).
Ullman, S.: Aligning pictorial descriptions: an approach to object recognition. Cognition 32(3) (1989) 193–254.
Ullman, S. and Basri, R.: Recognition by linear combinations of models. IEEE Trans. on Pattern Analysis and Machine Intelligence 13(10) (1991) 992–1006.
P. Van Hove: Model based silhouette recognition, Proc. of the IEEE Computer Society Workshop on Computer Vision (1987).
Weiss, I.: Projective invariants of shape, DARPA Image Unerstanding Workshop (1988) 1125–1134.
Weinshall, D.: Model-based invariants for 3D vision. International Journal on Computer Vision (1993).
Weinshall, D. and Basri, R.: Distance Metric between 3D Models and 2D Images for Recognition and Classification Proc. of IEEE conf. on Computer Vision and Pattern Recognition (1993) 220–225.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1994 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Basri, R. (1994). Recognition by combinations of model views: Alignment and invariance. In: Mundy, J.L., Zisserman, A., Forsyth, D. (eds) Applications of Invariance in Computer Vision. AICV 1993. Lecture Notes in Computer Science, vol 825. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-58240-1_23
Download citation
DOI: https://doi.org/10.1007/3-540-58240-1_23
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-58240-3
Online ISBN: 978-3-540-48583-4
eBook Packages: Springer Book Archive