• DocumentCode
    3010695
  • Title

    Region segmentation using edge based circle growing

  • Author

    Shimbashi, T. ; Kokubo, Y. ; Shirota, N.

  • Author_Institution
    Res. Center, Sony Corp., Tokyo, Japan
  • Volume
    3
  • fYear
    1995
  • fDate
    23-26 Oct 1995
  • Firstpage
    65
  • Abstract
    A new region segmentation method using edge based circle growing technique to construct texture regions for a natural image is proposed. This method can be used in object based coding for constructing closed regions to form contours. Also, it is robust against broken, noisy and isolated edges. Unlike the previous relaxation methods, e.g. the snake scheme, initialization parameters are not required to do region segmentation in this new method. Simulation results have shown that automatic region segmentation of the natural image can be achieved using this new method
  • Keywords
    edge detection; image coding; image segmentation; image texture; broken edges; closed regions; contour formation; edge based circle growing; isolated edges; natural image; noisy edges; object based coding; region segmentation; texture regions; Bit rate; Data mining; Discrete cosine transforms; Image coding; Image edge detection; Image processing; Image recognition; Image segmentation; Merging; Robustness;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Image Processing, 1995. Proceedings., International Conference on
  • Conference_Location
    Washington, DC
  • Print_ISBN
    0-8186-7310-9
  • Type

    conf

  • DOI
    10.1109/ICIP.1995.537581
  • Filename
    537581