• DocumentCode
    2089869
  • Title

    A structural approach to finding the point correspondences between two frames

  • Author

    Cheng, Jin-Chang ; Don, Hon-Son

  • Author_Institution
    Dept. of Electr. Eng., State Univ. of New York, Stony Brook, NY, USA
  • fYear
    1988
  • fDate
    24-29 Apr 1988
  • Firstpage
    1810
  • Abstract
    A recursive-descent tree-traversal algorithm is presented to find the point correspondences between two frames. Given two sets of noisy 3-D feature points on multiple rigid objects at two time-sequential views, the points at the first frame are constructed as a graph, and another graph is derived from those feature points at the second frame so that maximal matching points and minimum matching error are obtained. The correspondence of the vertices can then be found. The algorithm can also be used when the number of points at two frames are different. The computation time of the algorithm is large when the number of feature points is large. A date-set-splitting strategy for such cases is presented which can significantly reduce the computation time. Another algorithm is presented in which motion parameters are estimated from matched subgraphs and are used to guide the matching for the rest nodes. Computer simulations are performed to show the efficiency and accuracy of the algorithms
  • Keywords
    computer vision; graph theory; parameter estimation; computer vision; date-set-splitting strategy; graph theory; maximal matching points; minimum matching error; noisy 3-D feature points; parameter estimation; point correspondences; recursive-descent tree-traversal algorithm; Computer simulation; Computer vision; Data analysis; Matrix decomposition; Motion estimation; Noise reduction; Parameter estimation; Tree graphs;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Robotics and Automation, 1988. Proceedings., 1988 IEEE International Conference on
  • Conference_Location
    Philadelphia, PA
  • Print_ISBN
    0-8186-0852-8
  • Type

    conf

  • DOI
    10.1109/ROBOT.1988.12328
  • Filename
    12328