• DocumentCode
    2026277
  • Title

    Interactive object extraction using hierarchical graph cuts

  • Author

    Zhang, Yan ; Song, Haohao ; Gu, Jian ; Yu, Shengyang ; Yang, Jie

  • Author_Institution
    Third Res. Inst., Minist. of Public Security, Shanghai, China
  • fYear
    2010
  • fDate
    23-25 Nov. 2010
  • Firstpage
    851
  • Lastpage
    858
  • Abstract
    Foreground object extraction, which aims to accurately separate a foreground object from its background in still images, plays an important role in many computer vision applications. An interactive object extraction method by extending the graph cut approach is presented in this paper. The coarse-to-fine object segmentation method makes it efficiently to specify the main object and adjust the detail of object. The user interaction is simplified to drawing a rectangle around the desired object, followed by optional boundary editing. The coarse scale segmentation is performed on the basis of the initial interactive rectangle, the accurate boundary portrayal is performed at the finer scale. A pyramid structure provides a framework for the processing, and this hierarchical structure ensures rapid boundary mapping between pyramid levels. Experimental results on multiple kinds of color images show the effectiveness and convenience of the approach.
  • Keywords
    computer vision; feature extraction; graph theory; image segmentation; boundary editing; boundary mapping; boundary portrayal; coarse scale segmentation; coarse-to-fine object segmentation method; computer vision; foreground object extraction; hierarchical graph cut approach; interactive object extraction method; pyramid structure; Color; Equations; Image color analysis; Image resolution; Image segmentation; Mathematical model; Pixel;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Audio Language and Image Processing (ICALIP), 2010 International Conference on
  • Conference_Location
    Shanghai
  • Print_ISBN
    978-1-4244-5856-1
  • Type

    conf

  • DOI
    10.1109/ICALIP.2010.5685212
  • Filename
    5685212