• DocumentCode
    2458350
  • Title

    Stable Affine Frames on Isophotes

  • Author

    Perdoch, Michal ; Matas, Jiri ; Obdrzalek, Stepan

  • Author_Institution
    Czech Tech. Univ., Prague
  • fYear
    2007
  • fDate
    14-21 Oct. 2007
  • Firstpage
    1
  • Lastpage
    8
  • Abstract
    We propose a new affine-covariant feature, the stable affine frame (SAF). SAFs lie on the boundary of extremal regions, i.e. on isophotes. Instead of requiring the whole isophote to be stable with respect to intensity perturbation as in maximally stable extremal regions (MSERs), stability is required only locally, for the primitives constituting the three-point frames. The primitives are extracted by an affine invariant process that exploits properties of bitangents and algebraic moments. Thus, instead of using closed stable isophotes, i.e. MSERs, and detecting affine frames on them, SAFs are sought even on some unstable extremal regions. We show experimentally on standard datasets that SAFs have repeatability comparable to the best affine covariant detectors tested in the state-of-the-art report (Mikolajczyk et al., 2005) and consistently produce a significantly higher number of features per image. Moreover, the features cover images more evenly than MSERs, which facilitates robustness to occlusion.
  • Keywords
    covariance analysis; feature extraction; affine-covariant feature detection; algebraic moment; bitangent moment; intensity perturbation; maximally stable extremal region; stable affine frame; stable isophotes; Application software; Computer vision; Detectors; Object detection; Object recognition; Robustness; Shape; Signal processing; Stability; Testing;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Computer Vision, 2007. ICCV 2007. IEEE 11th International Conference on
  • Conference_Location
    Rio de Janeiro
  • ISSN
    1550-5499
  • Print_ISBN
    978-1-4244-1630-1
  • Electronic_ISBN
    1550-5499
  • Type

    conf

  • DOI
    10.1109/ICCV.2007.4408887
  • Filename
    4408887