Electrical Engineering
      and Computer Sciences

Electrical Engineering and Computer Sciences

COLLEGE OF ENGINEERING

UC Berkeley

Scale-space and edge detection using anisotropic diffusion

Pietro Perona and Jitendra Malik

EECS Department
University of California, Berkeley
Technical Report No. UCB/CSD-88-483
December 1988

http://www.eecs.berkeley.edu/Pubs/TechRpts/1988/CSD-88-483.pdf

The scale-space technique introduced by Witkin involves generating coarser resolution images by convolving the original image with a Gaussian kernel, or equivalently by using the original image as the initial condition of a diffusion process. This approach has a major drawback; it is difficult to obtain accurately the locations of the 'semantically meaningful' edges at coarse scales. In this paper we suggest a new definition of scale-space, and introduce a class of algorithms that realize it using a diffusion process. The diffusion coefficient is chosen to vary spatially in such a way as to encourage intra-region smoothing in preference to inter-region smoothing. It is shown that the 'No new maxima should be generated at coarse scales' property of conventional scale space is preserved. As the region boundaries in our approach remain sharp, we obtain a high quality edge detector which successfully exploits global information. Experimental results are shown on a number of images. The algorithm involves simple, local operations replicated over the image making parallel hardware implementation feasible.


BibTeX citation:

@techreport{Perona:CSD-88-483,
    Author = {Perona, Pietro and Malik, Jitendra},
    Title = {Scale-space and edge detection using anisotropic diffusion},
    Institution = {EECS Department, University of California, Berkeley},
    Year = {1988},
    Month = {Dec},
    URL = {http://www.eecs.berkeley.edu/Pubs/TechRpts/1988/5318.html},
    Number = {UCB/CSD-88-483},
    Abstract = {The scale-space technique introduced by Witkin involves generating coarser resolution images by convolving the original image with a Gaussian kernel, or equivalently by using the original image as the initial condition of a diffusion process. This approach has a major drawback; it is difficult to obtain accurately the locations of the 'semantically meaningful' edges at coarse scales. In this paper we suggest a new definition of scale-space, and introduce a class of algorithms that realize it using a diffusion process. The diffusion coefficient is chosen to vary spatially in such a way as to encourage intra-region smoothing in preference to inter-region smoothing. It is shown that the 'No new maxima should be generated at coarse scales' property of conventional scale space is preserved. As the region boundaries in our approach remain sharp, we obtain a high quality edge detector which successfully exploits global information. Experimental results are shown on a number of images. The algorithm involves simple, local operations replicated over the image making parallel hardware implementation feasible.}
}

EndNote citation:

%0 Report
%A Perona, Pietro
%A Malik, Jitendra
%T Scale-space and edge detection using anisotropic diffusion
%I EECS Department, University of California, Berkeley
%D 1988
%@ UCB/CSD-88-483
%U http://www.eecs.berkeley.edu/Pubs/TechRpts/1988/5318.html
%F Perona:CSD-88-483