From vision@deimos.ads.com Thu Oct 13 18:40:25 1988
Received: from ads.com by deimos.ads.com (5.59/1.11)
	id AA16455; Thu, 13 Oct 88 18:40:22 PDT
Received: from deimos.ads.com by ads.com (5.59/1.17)
	id AA23638; Thu, 13 Oct 88 18:39:25 PDT
Received: by deimos.ads.com (5.59/1.11)
	id AA16409; Thu, 13 Oct 88 18:34:13 PDT
Message-Id: <8810140134.AA16409@deimos.ads.com>
Date: 13 Oct 1988 18:11:26-PST
From: Vision-List moderator Phil Kahn <Vision-List-Request@ads.com>
Errors-To: Vision-List-Request@ads.com
Reply-To: Vision-List@ads.com
Subject: Vision-List delayed redistribution
To: Vision-List@ads.com
Status: RO

Vision-List Digest	Thu Oct 13 18:11:26 PDT 1988

 - Send submissions to Vision-List@ADS.COM
 - Send requests for list membership to Vision-List-Request@ADS.COM

Today's Topics:

 Circle Detection
 Need info on uses of Gabor Transform

----------------------------------------------------------------------

Date: Tue, 11 Oct 88 08:34:18 -0400
Subject: Circle Detection
From: "Kenneth I. Laws" <klaws@note.nsf.gov>


Another approach to circle detection is to compute circle centers from
pairs of edge vectors, then use cluster analysis.  This maintains
floating-point resolution without the overhead of a large Hough
accumulator, but may require multiple passes through the database of
hypothesized circle centers.  It's best to use only edge vectors at
approximate right angles, and to discard any pairs for which one edge
vector is further from the computed center than is the other.

I wrote a paper on this while I was at USC, back about 1977.
My papers are still in moving boxes, so I don't have this and
other citations handy.  There have been a number of circle or
ellipse finders based on Hough methods.  Some of the new neural
network approaches (such as adaptive resonance) could also be
used -- they're mostly just template matching, but the more
complex models can handle distortion and perhaps occlusion.

For work that depends on local continuity and gestalt grouping,
you might look into the work of Deborah Walters and her students
at SUNY.  They might not be able to recognize circles, but they
can pull out closed (or even open) curves.  Another source of
ideas is the engineering literature on finding curves in
handwriting and hand-drawn circuit diagrams or flowcharts.

					-- Ken Laws

------------------------------

Date: 12 Oct 88 01:33:07 GMT
From: efrethei@BLACKBIRD.AFIT.AF.MIL (Erik J. Fretheim)
Subject: Need info on uses of Gabor Transforms
Organization: Air Force Institute of Technology; WPAFB, OH


Has anyone seen the Gabor transform used for anything other than
edge/texture detection or data compression?  If you know of any other
interesting applications please let me know.

   Thank you for your time
	ejf

	[ Gabor functions have also been used for low-level motion/energy
	  detection.  Some references are Adelson and Bergen, J. Opt.
	  Soc. Amer. A, V2, N2, 2/85; Heeger, IEEE Proc. Intern. Conf.
	  Comp. Vision (ICCV), 6/87; Fleet and Jepson, Comput. Intell.,
	  V1, 1985.  The original paper by Gabor (who received the Nobel
	  prize for his work in holography) is "Theory of Communication,"
	  J. Inst. Electr. Eng., V93, 1946. 
			phil... 	]

------------------------------

End of VISION-LIST
********************

