Newsgroups: sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!udel!darwin.sura.net!usenews!news
From: lybanon@snaps.nrlssc.navy.mil (Matthew Lybanon)
Subject: Re: [Q] Morphological Operators
Message-ID: <D2yzz1.913@usenews.nrlssc.navy.mil>
X-Posted-From: InterNews 1.0.4@matt.nrlssc.navy.mil
Lines: 51
Sender: news@usenews.nrlssc.navy.mil
Nntp-Posting-Host: matt.nrlssc.navy.mil
Organization: U.S. Naval Research Laboratory
References:  <3g0hu9$sop@ccserver.cc.ic.ac.uk>
X-Authenticated: lybanon on POP host snaps.nrlssc.navy.mil
Date: Wed, 25 Jan 1995 16:47:23 GMT

In article <3g0hu9$sop@ccserver.cc.ic.ac.uk>
atae@spva.ph.ic.ac.uk (Ata Etemadi) writes:

> G'Day
> 
> Could someone point me to some good literature describing how
> one goes about designing a structuring element (the name used 
> by Haralick to describe the set used in conjunction with the 
> operator) for the opening and closing operators to achieve a 
> particular outcome ? I have read the Haralick paper in PAMI but 
> it only explains what these operators are and what effect a 
> given structuring element has after operating on an image. I
> am not even sure morphological operators are suited for solving 
> my particular problem.
> 
> Briefly, after trying to detect edges in an image I end up 
> with 'valleys' which are almost black in the middle and 
> climb to the value associated with the surrounding region
> within 2 to 3 pixels. What I want to do is 'thin' these valleys
> to 1 pixel thickness. Basically so I end up with just a thin
> black  line as opposed to a thick grayscale line. I thought 
> maybe morphological operators are the answer but I can't be 
> sure. Any help would be greatly appreciated. I will make a 
> summary of all responses and post them here. Thanks in advance.
> 
>         best regards
>                 Ata <(|)>.
> -- 
>          Dr Ata Etemadi, Blackett Laboratory,
>          Space and Atmospheric Physics Group,
>          Imperial College of Science, Technology, and Medicine,
>          Prince Consort Road, London SW7 2BZ, ENGLAND
> Internet/Arpanet/Earn/Bitnet: atae@spva.ph.ic.ac.uk
> Span                        :  SPVA::atae
> UUCP/Usenet                 :  atae%spva.ph.ic@nsfnet-relay.ac.uk


     One response to your request recommends Serra's book and other
publications by him and his group.  I agree that Serra is very good,
but not at all easy to read.  Haralick isn't much easier, although it
is a gold mine of information.  For some more information that you
might find somewhat less rough going, I recommend an issue of Computer
Vision, Graphics, and Image Processing that had a special section on
mathematical morphology (edited by Michael M. Skolnick):  volume 35,
number 3, September 1986.

     For an approach to edge detection based directly on mathematical
morphology (with the basic definitions somewhat modified), I recommend:
          S. Krishnamurthy et al., "Histogram-Based Morphological Edge
          Detector," IEEE Trans. on Geosci. and Rem. Sens., vol. 32,
no.
          4, pp. 759-767, July 1994.
The paper contains references to earlier work in this area.
