• DocumentCode
    2429281
  • Title

    Automatic recognition of exception points using polynomial expression

  • Author

    Hoshino, Takayuki ; Fujiwara, Takayuki ; Koshimizu, Hiroyasu ; Hatano, Yasuyo

  • Author_Institution
    Chukyo Univ., Toyota
  • fYear
    2007
  • fDate
    17-20 Oct. 2007
  • Firstpage
    1342
  • Lastpage
    1347
  • Abstract
    We have not yet completed the automatic recognition of facial parts by image processing. Even if once we get the edge image for the contour of facial parts, due to the existence of the noise, the results are quite frequently not sufficient both in local and global aspects. This is caused mainly by the lack the top-down mechanism for compensating the insufficient Bottom-Up mechanism of image processing. Then, in this paper, a curve fitting mechanism in the image processing is introduced as a top-down processing for enforcing the total performance of contour extraction of the facial parts. An experiment for extracting the contour line extraction of the jaw was done and some preliminary experimental results were presented in this paper, because jaw contour was especially hard to detect at the field test of EXPO2005 Exhibition. Furthermore the possibility of the new deformation basis for the facial caricature is prospected so that the parameter space of the analytical curve could be introduced.
  • Keywords
    curve fitting; edge detection; face recognition; feature extraction; polynomials; automatic exception point recognition; curve fitting mechanism; edge image; facial part contour extraction; image processing; jaw contour; polynomial expression; top-down processing; Automatic control; Automation; Control systems; Curve fitting; Face recognition; Image processing; Image recognition; Polynomials; Skin; Testing; Extracting jaw contour; Facial caricature; Liner least square method; Top-down processing;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Control, Automation and Systems, 2007. ICCAS '07. International Conference on
  • Conference_Location
    Seoul
  • Print_ISBN
    978-89-950038-6-2
  • Electronic_ISBN
    978-89-950038-6-2
  • Type

    conf

  • DOI
    10.1109/ICCAS.2007.4406546
  • Filename
    4406546