Newsgroups: sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!gatech!howland.reston.ans.net!ix.netcom.com!netcom.com!sujo
From: sujo@netcom.com (Suresh Krishnajois)
Subject: [HELP}Edge detection, segmentation
Message-ID: <sujoD5rw1H.D43@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
X-Newsreader: TIN [version 1.2 PL1]
Date: Tue, 21 Mar 1995 04:20:04 GMT
Lines: 32
Sender: sujo@netcom17.netcom.com

Hi,

I'm looking to do the following, on a rather urgent (2 weeks) time frame:

1. Scan an image guaranteed to have a single object on a white background.
2. Detect the edge/perimeter/outline of this one object.
3. Throw away all background and interior pixels, retain just the edge.
4. Segment the edge pixel array and fit an appropriate number of polynomial
   curves.
5. Find the curvature of the edge at each edge pixel and sort the edge 
   pixels by ascending order of absolute value of curvature.
6. Also in the edge pixel array, identify points where 
   - curvature of the edge exeeds a specified threshold
   - curvature changes sign (i.e., edge goes from concave->convex or reverse)
7. Output the following data:
   - edge pixel coordinates
   - coefficientss of the polynomials that segment the edge
   - coordinates of the points per the curvature criteria in step 6 above.
   - edge pixels sorted by descending order of absolute value of curvature
     of the edge, derived in step 5 above

Is this something way out or eminently doable ? Is there public domain 
source code available which I can use and modify for non-commercial use ?

I'd immensely appreciate any pointers on this. Thanks in advance.

- Suresh
-- 
---------------------------------------------------------
Suresh Krishna Jois | sujo@netcom.com | Cyberspace Gypsy. 
Traceroute: ->Informix->SiliconGraphics->Sun Microsystems
            ->Home Shopping Network/Internet Subsidiary
Virtual Camp: http://www.internet.net/People/Suresh
---------------------------------------------------------
