• DocumentCode
    3457624
  • Title

    Three Dimensional Face Recognition Based On Geodesic Distance

  • Author

    Chang, Jun-Yan ; Da, Fei-Peng ; Cai, Liang

  • Author_Institution
    Dept. of Autom., Southeast Univ., Nanjing, China
  • fYear
    2010
  • fDate
    21-23 Oct. 2010
  • Firstpage
    1
  • Lastpage
    5
  • Abstract
    Expression changes, especially the mouth open will bring errors to the calculation of geodesic distance using classical method. In order to address the issue, an novel approach to distance calculation between points based on geodesic is presented in this paper. Firstly, a square is obtained by fitting four key points around the mouth. If the projected line between two marks is intersected with the square, bounding box method is used to determine the crossing points. Then we respectively compute the distances between the crossing and the marks which will be added as the distance between land marks. Otherwise, FMM (Fast Marching Method) is directly applied. Distance matrix among the marks is extracted as feature to identify individual. Experiment on FRGC V2.0 shows that our modified method is more accurate and robust to expression.
  • Keywords
    differential geometry; face recognition; feature extraction; matrix algebra; FRGC V2.0; distance matrix; fast marching method; fetaure extraction; geodesic distance; square bounding box method; three dimensional face recognition; Face recognition; Feature extraction; Fitting; Iterative closest point algorithm; Mouth; Robustness; Three dimensional displays;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Pattern Recognition (CCPR), 2010 Chinese Conference on
  • Conference_Location
    Chongqing
  • Print_ISBN
    978-1-4244-7209-3
  • Electronic_ISBN
    978-1-4244-7210-9
  • Type

    conf

  • DOI
    10.1109/CCPR.2010.5659227
  • Filename
    5659227