Communications
Joint Invariants of a Triplet of Coplanar Conics: Stability and Discriminating Power for Object Recognition

https://doi.org/10.1006/cviu.1998.0617Get rights and content

Abstract

Joint invariants of a pair of coplanar conics have played a central role in the initial work on the application of invariants to object recognition. In this paper, we are concerned with the invariants of a triplet of coplanar conics. It will be shown that, in addition to the invariants associated to all pairs of three conics, there exists one new invariant which is only associated to the triplet of conics. This new invariant of the three conics can therefore be used to discriminate triplets of conics which might be pair-wise similar. All joint invariants of a triplet of coplanar conics will first be derived based on the invariant algebra of a net of ternary quadratic forms. Then, experimentations for discriminating power, accuracy, and stability of joint invariants are conducted both for simulated and real images. Finally, it is also shown that the method developed for a triplet of coplanar conics can be extended to any number of coplanar conics.

References (21)

  • F.L. Bookstein

    Fitting conic sections to scattered data

    Comput. Graphics Image Process.

    (1979)
  • S. Carlsson

    Projectively invariant decomposition of planar shapes

    Geometric Invariance in Computer Vision

    (1992)
  • O. Faugeras, What can be seen in three dimensions with an uncalibrated stereo rig?, Proceedings of the 2nd European...
  • D. Forsyth et al.

    Invariant descriptors for 3D object recognition and pose

    IEEE Trans. Pattern Anal. Machine Intell.

    (1991)
  • P. Gros, L. Quan, Projective invariants for vision, Grenoble, France, Dec....
  • P. Gros et al.

    3D projective invariants from two images

    Proceedings of the SPIE Conference on Geometric Methods in Computer Vision II

    (1993)
  • R. Hartley et al.

    Stereo from uncalibrated cameras

    Proceedings of the Conference on Computer Vision and Pattern Recognition

    (1992)
  • K. Kanatani

    Geometric Computation for Machine Vision

    (1993)
  • R. Mohr et al.

    Relative 3D reconstruction using multiple uncalibrated images

    Int. J. Robotics Res.

    (1995)
  • J. Mundy, D. Kapur, S. Maybank, P. Gros, L. Quan, Geometric interpretation of joint conic invariants, Geometric...
There are more references available in the full text version of this article.

Cited by (10)

  • Planar shape representation and matching under projective transformation

    2011, Computer Vision and Image Understanding
    Citation Excerpt :

    The conic basis as it is defined in this paper takes the non-parametric equation of conic as its basis and is realized by fitting piecewise conics to object’s enclosing contour which defines its shape. A similar approach has been reported previously in [36,37], where the authors used a specific number of conics to represent object contour and establish correspondences between the conics to compute a matching score. Former conic based methods, however, have two major limitations.

  • Image analysis and computer vision: 1998

    1999, Computer Vision and Image Understanding
  • Lunar Crater Identification in Digital Images

    2021, Journal of the Astronautical Sciences
View all citing articles on Scopus

J. MundyA. Zisserman

View full text