VISION-LIST Digest 1988 12 02

eZine lover (@eZine)

Published in 

VISION LIST Digest

 · 10 months ago

Vision-List Digest	Fri Dec 02 16:16:37 PDT 88 

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

Today's Topics: 

 DeScreening 
 dissemination of sharware for Image Processing and Computer Vision 
 Robotics Seminar 

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

Date:         Thu, 24 Nov 88 15:47:37 IST 
From: Shelly Glaser  011 972 3 5450119 <GLAS%TAUNIVM.BITNET@CUNYVM.CUNY.EDU> 
Subject:      DeScreening 

Please publish the following question in the Vision Newsletter: 

     I  am looking  for information  on practical  solutions to  the 
     "de-screening" problem: taking a half-toned image (like the one 
     in printed book or magazine)  and removing the half-tone screen 
     so we get a true continuous-gray-scale image (as opposed to the 
     binary  pulse area  modulated  half-tone  image).  The  obvious 
     solution, low-pass filtering, often kills  too much of the fine 
     details  in the  image,  so  I am  looking  for something  more 
     sophisticated. 

     Many thanks, 
                             Sincerely Yours, 
                                                       Shelly Glaser 

                    Department of Electronic, Communication, Control 
                                                and Computer Systems 
                                              Faculty of Engineering 
                                                 Tel-Aviv University 
                                                    Tel-Aviv, Israel 

                                                   FAX: 972 3 419513 
                               Computer network: GLAS@TAUNIVM.BITNET 

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

Date: 26 Nov 88 18:58:00 GMT 
From: annala%neuro.usc.edu@oberon.usc.edu (A J Annala) 
Subject: possible use of comp.ai.vision 
Organization: University of Southern California, Los Angeles, CA 

 
There has been some discussion in comp.graphics about using comp.ai.vision 
as the home for discussions about andf distribution of image processing 
software.  I personally suspect that this would not be an appropriate use 
of the comp.ai.vision group; however, I would appreciate email to my user 
account (which I will summarize) on this issue. 

Thanks, AJ Annala ( annala%neuro.usc.edu@oberon.usc.edu ) 

[ Discussions on IP sofware are most definitely appropriate for the Vision 
  List and comp.ai.vision.  Yet, as with other SIG networks, it is not  
  appropriate to submit the code in this forum.  Rather, if there is  
  shareware IP and CV software which should be disseminated, then a 
  new network newsgroup entitled something like COMP.BINARIES.VISION 
  should be established.  This requires a site and moderator for this new 
  net which can establish and manage this new facility.  Any volunteers? 

			phil...  	] 

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

Date: Tue, 29 Nov 88 19:30:46 PST 
From: binford@Boa-Constrictor.Stanford.EDU.stanford.edu (Tom Binford) 
Subject: Robotics Seminar 

			 Robotics Seminar 
			Monday, Dec 7, 1988 
		4:15pm Cedar Hall Conference Room 

 
            SOLVING THE STEREO CONSTRAINT EQUATION 

		Stephen Barnard 
		Artificial Intelligence Center 
		SRI International 

The essential problem of stereo vision is to find a disparity map 
between two images in epipolar correspondence. The stereo constraint 
equation, in any of its several forms, specifies a function of disparity 
that is a linear combination of photometric error and the first order 
variation of the map.  This equation can also be interpreted as the 
potential energy of a nonlinear, high dimensional dynamic system.  By 
simulating either the deterministic newtonian dynamics or the 
statistical thermodynamics of this system one can find approximate 
ground states (i.e., states of minimum potential energy), thereby 
solving the stereo constraint equation while constructing a dense 
disparity map. 

The focus of this talk will be a stochastic method that uses a 
microcanonical version of simulated annealing.  That is, it explicitly 
represents the heat in the system with a lattice of demons, and it cools 
the system by removing energy from this lattice.  Unlike the 
``standard'' Metropolis version of simulated annealing, which simulates 
the canonical ensemble, temperature emerges as a statistical property of 
the system in this approach.  A scale-space image representation is 
coupled to the thermodynamics in such a way that the higher frequency 
components come into play as the temperature decreases.  This method has 
recently been implemented on a Connection Machine. 

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

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