Pablo Arbelaez and Michael Maire and Charless Fowlkes and Jitendra Malik

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2010-17

February 16, 2010

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-17.pdf

This paper investigates two fundamental problems in computer vision: contour detection and image segmentation. We present state-of-the-art algorithms for both of these tasks. Our contour detector combines multiple local cues into a globalization framework based on spectral clustering. Our segmentation algorithm consists of generic machinery for transforming the output of any contour detector into a hierarchical region tree. In this manner, we reduce the problem of image segmentation to that of contour detection. Extensive experimental evaluation demonstrates that both our contour detection and segmentation methods significantly outperform competing algorithms. The automatically generated hierarchical segmentations can be interactively refined by userspecified annotations. Computation at multiple image resolutions provides a means of coupling our system to recognition applications.

---

BibTeX citation:

@techreport{Arbelaez:EECS-2010-17,
    Author= {Arbelaez, Pablo and Maire, Michael and Fowlkes, Charless and Malik, Jitendra},
    Title= {Contour Detection and Hierarchical Image Segmentation},
    Year= {2010},
    Month= {Feb},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-17.html},
    Number= {UCB/EECS-2010-17},
    Abstract= {This paper investigates two fundamental problems in computer vision: contour detection and image segmentation. We present state-of-the-art algorithms for both of these tasks. Our contour detector combines multiple local cues into a globalization framework based on spectral clustering. Our segmentation algorithm consists of generic machinery for transforming the output of any contour detector into a hierarchical region tree. In this manner, we reduce the problem of image segmentation to that of contour detection. Extensive experimental evaluation demonstrates that both our contour detection and segmentation methods significantly outperform competing algorithms. The automatically generated hierarchical segmentations can be interactively refined by userspecified
annotations. Computation at multiple image resolutions provides a means of coupling our system to recognition applications.},
}

EndNote citation:

%0 Report
%A Arbelaez, Pablo 
%A Maire, Michael 
%A Fowlkes, Charless 
%A Malik, Jitendra 
%T Contour Detection and Hierarchical Image Segmentation
%I EECS Department, University of California, Berkeley
%D 2010
%8 February 16
%@ UCB/EECS-2010-17
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-17.html
%F Arbelaez:EECS-2010-17