Ellipse based stereo vision

  • Johannes Buurman
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 588)


We propose a new stereo vision algorithm for finding circles in a scene. In both 2-D images, ellipses are found. The ellipses are matched in order to find circles in 3-D space. The method does not require a special camera alignment, instead both camera matrices must be known. Some results are presented, showing that the method is sufficiently fast and accurate for object recognition. After edge detection, a few seconds of CPU time are sufficient to find full circles with standard deviations of the order of 1–2% of the radius of the circles.


  1. [BB1]
    D. H. Ballard and C. M. Brown, Computer Vision, Prentice-Hall, Englewood Cliffs, New Jersey, (1982).Google Scholar
  2. [Bu1]
    J. Buurman, The Diac Object Recognition System, to be presented at SPIE conference on Applications of Artificial Intelligence X: Machine Vision and Robotics, Orlando (1992).Google Scholar
  3. [Bu2]
    J. Buurman, Ellipse based stereo vision. Internal report. Pattern Recognition group, faculty of Applied Physics, Delft University of Technology, (1992).Google Scholar
  4. [FS1]
    N.J. Foster and A.C. Sanderson, Determining Object Orientation Using Ellipse Fitting, SPIE Intelligent Robots and Computer Vision, vol. 521, (1984) 34–43.Google Scholar
  5. [Ho1]
    B.K.P. Horn, Robot Vision, McGraw-Hill, New York, (1986).Google Scholar
  6. [PP1]
    S.B. Pollard, J. Porrill, and J.E.W. Mayhew, Recovering partial 3D wire frames descriptions from stereo data, Image and Vision Computing, vol. 9, no. 1, (1991) 58–65Google Scholar
  7. [RG1]
    C.J. Rijnierse and F.C.A. Groen, Graph construction and matching for 3D object recognition, in: Pattern recognition and artificial intelligence — towards an integration, ed. L.N. Kanal, North Holland, Amsterdam, (1988)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1992

Authors and Affiliations

  • Johannes Buurman
    • 1
  1. 1.Pattern Recognition Group, Faculty of Applied PhysicsDelft University of TechnologyDelftThe Netherlands

Personalised recommendations