Measures for silhouettes resemblance and representative silhouettes of curved objects

  • Yoram Gdalyahu
  • Daphna Weinshall
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1065)


We claim that the task of object recognition necessitates a measure for image likelihood, that is: a measure for the probability that a given image is obtained from a familiar (pre-investigated) object. Moreover, in a system where objects are represented by 2D images, the best performance is achieved if those images are selected according to a maximum likelihood principle. This is equivalent to maximum stability of the image, or minimal change under a viewpoint perturbation. All of those qualities involve a quantitative comparison of similarity between images. We propose different metric functions which can be imposed on the image space of curved three dimensional objects. We use these metrics to detect the representative views (most stable and most likely views) of three test models. We find the same representative views under all the investigated metrics, suggesting that local maxima of stability and likelihood are metric independent. Our method of image comparison is based solely on the appearance of the occluding contour, hence our method is suitable for object recognition from silhouettes.


Feature Point Object Recognition Image Space Interest Point Dimensional Object 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    E. M. Arkin, L. P. Chew, D. P. Huttenlocher, K. Kedem, and J. S. B. Mitchell. An efficiently computable metric for comparing polygonal shapes. T-PAMI, pages 209–216, 1991.Google Scholar
  2. 2.
    J. Ben-Arie. The probabilistic peaking effect of viewed angles and distances with application to 3-d object recognition. T-PAMI, pages 760–774, 1990.Google Scholar
  3. 3.
    A. M. Bruckstein, R. J. Holt, A. N. Netravali, and T. J. Richardson. Invariant signature for planar shape recognition under partial occlusion. CVGIP: Image Understanding, pages 49–65, 1993.Google Scholar
  4. 4.
    A. M. Bruckstein, N. Katzir, M. Lindenbaum, and M. Porat. Similarity invariant recognition of partially occluded planar curves and shapes. IJCV, pages 271–285, 1992.Google Scholar
  5. 5.
    J. B. Burns, R. S. Weiss, and E. M. Riseman. View variation of point set and line segment features. T-PAMI, pages 51–68, 1993.Google Scholar
  6. 6.
    S. Duvdevani-Bar and S. Edelman. On similarity to prototypes in 3d object representation. Technical Report CS-TR 95-11, Weizmann Institute, July 1995.Google Scholar
  7. 7.
    W. T. Freeman. Exploiting the generic view assumption to estimate scene parameters. In proc. of ICCV, pages 347–356, 1993.Google Scholar
  8. 8.
    D. P. Huttenlocher and S. Ullman. Object recognition using alignment. In proc. of ICCV, pages 102–111, 1987.Google Scholar
  9. 9.
    N. Katzir, M. Lindenbaum, and M. Porat. Planar curve segmentation for recognition of partially occluded shapes. In proc. of ICPR, 1990.Google Scholar
  10. 10.
    D. Keren, D. Cooper, and J. Subrahmonia. Describing complicated objects by implicit polynomials. T-PAMI, pages 38–53, 1994.Google Scholar
  11. 11.
    Y. Lamdan, J. T. Schwartz, and H. J. Wolfson. Affine invariant model based object recognition. T-RA, pages 578–589, 1990.Google Scholar
  12. 12.
    P. J. van Otterloo. A Contour-Oriented Approach to Shape Analysis. Prentice Hall International, 1991.Google Scholar
  13. 13.
    D. Weinshall and M. Werman. Disambiguation techniques for recognition in large databases and for under-constrained reconstruction. In proc. of Intr. Symp. on Comp. Vision, pages 425–430, Coral Gables, USA, 1995.Google Scholar
  14. 14.
    D. Weinshall, M. Werman, and N. Tishby. Stability and likelihood of views of three dimensional objects. Computing Suppl., 11:237–256, 1996.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1996

Authors and Affiliations

  • Yoram Gdalyahu
    • 1
  • Daphna Weinshall
    • 1
  1. 1.Institute of Computer ScienceThe Hebrew UniversityJerusalemIsrael

Personalised recommendations