• DocumentCode
    2445220
  • Title

    Salient Object Contour Detection Based on Boundary Similar Region

  • Author

    Jin Zhan ; Bo Hu

  • Author_Institution
    State-Province Joint Lab. of Digital Home Interactive Applic., Sun Yat-sen Univ., Guangzhou, China
  • fYear
    2012
  • fDate
    23-25 Nov. 2012
  • Firstpage
    335
  • Lastpage
    339
  • Abstract
    Salient object contours are important cues for object detection and shape recognition in computer vision. We propose a novel contour detection method for salient object considering salient edges operator. The method traces continuous boundaries from salient edges, computes similar regions between boundaries with direction channel and distance by Boundary Similar Region (BSR) algorithm. Then we fill the boundary similar regions by linear interpolation method to ensure that boundaries belong to a same contour. We demonstrate our method applied to salient object segmentation. Experiments indicate that our method show better accuracy of salient object boundary. On the basis of edge manipulation, our method is compact and simple that performs efficiently.
  • Keywords
    computer vision; edge detection; image segmentation; interpolation; object detection; shape recognition; BSR algorithm; boundary similar region algorithm; computer vision; direction channel; direction distance; linear interpolation method; object detection; salient edges operator; salient object boundary accuracy; salient object contour detection method; salient object segmentation; shape recognition; Accuracy; Computer vision; Educational institutions; Filling; Image edge detection; Interpolation; Object segmentation; Oriented Chamfer Matching; boundary similar region algorithm; contour detection; salient object segmentation;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Digital Home (ICDH), 2012 Fourth International Conference on
  • Conference_Location
    Guangzhou
  • Print_ISBN
    978-1-4673-1348-3
  • Type

    conf

  • DOI
    10.1109/ICDH.2012.74
  • Filename
    6376435