[comp.ai.vision] VISION-LIST digest 10.13

Vision-List-Request@ADS.COM (Vision-List moderator Phil Kahn) (03/23/91)

VISION-LIST Digest    Fri Mar 22 13:05:16 PDT 91     Volume 10 : Issue 13

 - Send submissions to Vision-List@ADS.COM
 - Send requests for list membership to Vision-List-Request@ADS.COM
 - Access Vision List Archives via anonymous ftp to ADS.COM

Today's Topics:

 Need ground truth stereo disparity imagery
 Recognition of Moving Objects?
 Optimization in low-level image processing, Recurrent Nets.
 Features Extraction
 Thinning/skeletonizing Algorithms
 CAD databases in vision systems
 I need NITF Source Code
 Video WORM Summary

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

Date: Tue, 19 Mar 91 22:14:15 GMT
From: tolley@eola.cs.ucf.edu (Derek Tolley)
Subject: Need ground truth stereo disparity imagery

	We are interested in finding disparity maps with ground truth
for some stereo pairs. We are trying to compare the accuracy of some
stereo algorithms, and would like to have a verified disparity map to
compare with the results. If anyone has the sites where these pairs
and their disparity maps can be found or other information, please
email me.  Thank you.

			Derek Tolley (tolley@eola.cs.ucf.edu)
[ Such imagery would be a good addition to the Vision List archives.
						phil...  ]

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

Date: Mon, 18 Mar 91 14:11:57 EST
From: Nigel.Goddard@B.GP.CS.CMU.EDU
Subject: Recognition of Moving Objects?

Fellow visionaries,

I am interested in finding out about any work on identifying moving
objects, particularly non-rigid articulated objects such as humans
walking, running, including recognition of the movement.  I am aware
of structure from motion work that recovers the structure of the object
from the motion, but I know of no work that goes on to model and identify
complex movements such as walking or running.  I will provide a summary
of responses if it is requested.

Nigel Goddard
goddard@b.gp.cs.cmu.edu

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

Date: Thu, 14 Mar 91 08:37:25 GMT
From: eba@computing-maths.cardiff.ac.uk (Eduardo Bayro)
Organization: Univ. of Wales Coll. of Cardiff, Dept. of Electronic & Systems Engineering
Subject: Optimization in low-level image processing, Recurrent Nets.

Hi friends !
Many optimization problems in low level image processing could be
tackle with  recurrent or recursive neural network implementations.
Is anybody outthere who can send to me references and comments?
Applications include filtering, enhancement, segmentation and feature
extraction.  The only paper I have got is:
D. Suter " Neural Net Surface Interpolation " 1987 Int'l Conf. Systems, Man
and Cybernetics, Alexandria, VA, pp. 118-123, Aug. 1988. 
I wil resume them and post them back. Thanking you in advance.
Eduardo.

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

Date: Thu, 14 Mar 91 09:09:59 EST
From: skuo@bach.rutgers.edu
Subject: Features Extraction

We currently start a project related to image recognition/understanding,
Character Recognition is one of tasks.
It will be highly appreciated if anyone can provide references or pointers
to Feature Extraction algorithms.
We will post the summary if any response is available.
Thank you very much in advance.

S. Kuo
Email: skuo@caip.rutgers.edu

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

Date: Mon, 18 Mar 91 10:42:18 EST
From: skuo@bach.rutgers.edu
Subject: Thinning/skeletonizing Algorithms

I am working on an image recognition project.  Before recognizing the
object, I would like to skeletonize the image of objects.  Thus, it
will be highly appreciated if anyone can give me some references
and/or pointers about the skeletonizing/thinning algorithms.  I will
post a summary of all responses.  Thank you very much in advance.

S. kuo
skuo@caip.rutgers.edu

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

Date: 19 Mar 91 05:29:16 GMT
From: marcr@math.mscs.mu.edu (Marc R.)
Organization: Marquette University - Milwaukee, Wisconsin
Subject: CAD databases in vision systems

I am looking for some good (or even bad) litature on the databases used 
in vision software used in CAD solid/wireframe modeling.

I've gotten the headder from BRL-CAD's database, and would like to have
more of said headders for this little project.   So if you are:
1) A CAD vendor
2) A database vendor in the Cad world
3) into vision (machine that is)
4) read an article which might pretain
5) just plain informed
Please send me e-mail, as the only group I read at the moment is
comp.databases.

Marc Rassbach     marcr@marque.mu.edu         If you take my advice, that
...marque!msoe3b2!milestn!marc               is your problem, not mine!
              If it was said on UseNet, it must be true.
   Unix - It's a nice place to live, but you don't want to visit there.

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

Date: 13 Mar 91 19:53:52 GMT
From: barboni@sbcs.sunysb.edu (Carol Barboni)
Organization: State University of New York at Stony Brook
Subject: I need NITF Source Code

I am looking for source code for the National Image Transmission
Format (NITF); in particular, I am interested in obtaining the
complete source code for the functionality which translates _from_
the NITF standard format _into_ a form acceptable to UNIX platforms.
My hardware suite is SUN-based.  Thanks in advance.

Carol R. Barboni, Grumman Data Systems, (516) 682-8355
barboni@gdstech.grumman.com

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

Date: Mon, 18 Mar 91 16:46:09 PST
From: spock@windchime.arc.nasa.gov (Son of Sarak)
Subject: Video WORM Summary

Hello,

After posting a request for information about video WORM drives, I
received several responses, many of which were posted directly to the
Vision-List.  Apparently, there are two manufacturers of video WORM
drives: Sony and Panasonic.  Since the responses contained much
greater depth of information about the Sony models, I have waited to
post these results until I was able to include equivalent information
about the Panasonic models.

I. Sony CRV (Component Recording Video) System

   Introduction:  
	- The basic record/playback system consists of 2 units.
	  1. LVS-5000 processor and controller unit
	  2. LVR-5000 recorder/player mechanical drive unit
	- The LVS-5000 can control up to 10 of the LVR-5000 units,
	  allowing the discs to be recorded/played consecutively.
	- A stand-alone player-only unit, the LVA-7000, is also available.

   Physical Characteristics:
	Model 			Size			Weight
	LVS-5000	375(W)x180(H)x450(D)mm		12.5 kg
	LVR-5000	375(W)x180(H)x530(D)mm		18   kg
	LVA-7000	424(W)x213(H)x530(D)mm		22   kg

   Video Formats (input):
	- NTSC composite, RGB, R-Y/Y/B-Y
	- The system can also be set to record only black and white
	  RS-170 video with an increased bandwidth.
	- PAL is supported with an optional board.

   Resolution/Bandwidth (luminance):	NTSC		4.5 MHz
					RS-170		6.7 MHz

   S/N ratio:				48 dB

   Media:				12" disc (double sided only?)

   Media Capacity:			24 minutes or 43,500 frames per side

   Seek Time:				0.5 seconds (full stroke average)

   Computer Control:			RS-232C		(LVR-5000)
					RS-232C, RS-422 (LVA-7000)

   Approximate Cost:			all models US$8000 
					This is the approximate cost for 
					government agencies, list price 
					is probably 20-25% higher.

II. Panasonic TQ-303xF Series

   Introduction:
	- The Panasonic models are all stand-alone units.  The new
	  TQ-303xF series models supersede the earlier TQ-2026F unit
	  which was mentioned in several responses from the network.
	- The Panasonic models are
	  TQ-3031F	recorder/player
	  TQ-3032F	player only
	  TQ-3038F	monochrome recorder/player

   Physical Characteristics:	
	size	430(W)x155(H)x546(D)mm
	weight	18 kg

   Video Formats (input):	
	TQ-3031F	NTSC composite, RGB, S-video(Y/C)
	TQ-3038F	RS-170

   Horizontal Resolution:
	TQ-3031F	380 TV-lines	(Normal mode)
			450 TV-lines	(Hi-Res mode)
	TQ-3038F	600 TV-lines

   S/N ratio:			45 dB

   Media:			12" disc (single or double sided)

   Media Capacity:	
	TQ-3031F	30 minutes (54,000 frames) per side	(Normal mode)
			20 minutes (36,000 frames) per side	(Hi-Res mode)
	TQ-3038F	20 minutes (36,000 frames) per side

   Seek Time:			0.7 second (average)

   Computer Control:
	RS-232C	 standard equipment
	RS-422   available as option for TQ-3031F/TQ-3032F

   Approximate Cost:	TQ-3031F/TQ-3038F  	US$20,000
			TQ-3032F		US$5,000 

It should be noted that all of the models described have numerous
features for playing/recording still frames and continuous video.
Also, all models have provisions for storing some alphanumeric data
directly on the discs for picture indices, etc.

In addition to the WORM drives discussed above, Panasonic is
introducing a Rewritable Optical Disc Recorder, the LQ-4000.  The unit
is slightly taller and heavier than the TQ-3031F, and accepts
R-Y/Y/B-Y input in addition to the other formats accepted by the
TQ-3031F.  In other respects, the LQ-4000 specifications appear
identical to those of the TQ-3031F, including the fact that it
operates in both Normal and Hi-Res modes.

Thanks to those who took the time and effort to respond to my request.

				Phillip Smith
				pnsmith@ames.arc.nasa.gov

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

End of VISION-LIST digest 10.13
************************