Copy Link
Add to Bookmark
Report
VISION-LIST Digest 1990 05 18
Vision-List Digest Fri May 18 16:09:21 PDT 90
- Send submissions to Vision-List@ADS.COM
- Send requests for list membership to Vision-List-Request@ADS.COM
Today's Topics:
Mandrill picture available
Re: Need for structured light images
Research Associate posts in Neural Networks and Computer Vision
IBM/Metro Toronto Police image processing
TR on Tracking and Shape
Bibliographic info
Re: Other image processing/vision-related lists.
Updated info on the Vision List and related services
----------------------------------------------------------------------
Date: Fri, 18 May 90 09:15:09 EDT
From: dfausett@zach.fit.edu ( Donald W. Fausett)
Subject: Mandrill picture available
Status: RO
Phil:
Here is one response that looks encouraging. Maybe the mandrill image
can be the start of your collection. Thanks for your assistance. -- Don
[ The mandrill image gets the dubious distinction of being the first
imagery made available through the Vision List archive. I would very
much like to add other imagery which is of general interest to the
community. Please contact me at Vision-List-Request@ADS.COM if you
have contributions. See info at the end of this message.
phil... ]
From: pkube@UCSD.EDU (Paul Kube) [ The provider of the mandrill images ]
There are three files, one for each of the red, green, and blue
components:
-rw-rw-r-- 1 ruzzo 262176 May 17 16:24 mandrill.blue
-rw-rw-r-- 1 ruzzo 262176 May 17 16:24 mandrill.green
-rw-rw-r-- 1 ruzzo 262176 May 17 16:24 mandrill.red
Each file is in Sun rasterfile format, colormap type RMT_NONE,
maplength 0, height 512, width 512, depth 8. For those without Sun
documentation, this means that each file starts with a 32-byte header,
followed by 512x512 bytes of image data, one byte per pixel, in row
major order.
I have not compressed the files, because the Lempel-Ziv algorithm
implementation in standard Unix compress is unable to. The adaptive
Huffman technique in old Unix compact can reduce their size by only
a few per cent. This is a difficult image for standard compression
techniques because of large regions of uncorrelated signal on each scan
line.
Enjoy,
Paul Kube@ucsd.edu
------------------------------
Date: Fri, 18 May 90 14:00:35 EDT
From: "Prof. Jeff Labuz" <labuz@vision1.ece.scarolina.edu>
Subject: Re: Need for structured light images
This is in response to a recent request for a calibrated sequence
of dynamic images of 3D surfaces illuminated by structured light.
You may be able to use the projector/camera simulator I described
in the article "Three-dimensional surface mapping simulator: theory,
capabilities and operation," in the British journal IMAGE AND VISION
COMPUTING, vol. 3, no. 1, Feb. 1985, pp. 36-39. The simulator has
several advantages over real data for proof-of-concept experiments.
For example, the calibration matrices for the pinhole-modeled camera
and projector are created by the programmer, so that any desired
rigid (6 DOF) motion can be incorporated. The projection pattern can
be customized to suit the algorithm. And the 3-D surface is modeled
as any possible quadric (second order) surface, although higher-order
surfaces could be handled, and it can be rigidly displaced as
desired between frames. Unfortunately, there are typographical
errors in equations (3a), (3b), and (4), and step (13) of the
algorithm on page 38 should end in GO TO (2). I can send a copy of
the entire article, or just the corrections to the three equations,
to those who are interested.
Jeffrey Labuz, Dept of Elec and Comp Engr
Univ of South Carolina, Columbia, SC 29208
E-mail: labuz@vision1.ece.scarolina.edu
------------------------------
Date: 18 May 1990 13:15:35 GMT
From: austin%minster.york.ac.uk@NSFnet-Relay.AC.UK
Subject: Research Associate posts in Neural Networks and Computer Vision
University of York
Advanced Computer Architecture Group
Department of Computer Science
York
UK
Two Research Associate posts in
Neural Networks and Computer Vision
Applications are invited for two, three year research asso-
ciateships within the department of Computer Science on a
DTI/SERC JFIT research project. Applicants should preferably
have programming and research experience of computer vision
and neural networks.
The work is the domain of automated visual guidance of air-
borne vehicles. In particular, in the development of paral-
lel vision algorithms, based on neural networks. A major aim
is to develop methods for combining high and low level
information in the recognition process.
The project is in collaboration with the University of
Surrey and Rutherford Appleton laboratories and, as indus-
trial partners, RSRE Malvern and British Aerospace plc.
Salaries will be on the 1A scale (10,458 --- 12,381 pounds
sterling). Informal enquiries may be made to Dr. Jim Austin
(UK 0904 432734,email: austin@uk.ac.york.minster). Further
particulars may be obtained from The Registrar's Department,
University of York, Heslington, York, YO1 5DD, UK. to whom
three copies of a curriculum vitae should be sent. The clos-
ing date for applications is 7 june 1990. Please quote
reference number 6303.
------------------------------
Date: 18 May 90 20:07:42 GMT
From: king@ug.cs.dal.ca (Yossarian Yggy King)
Subject: IBM/Metro Toronto Police image processing
Organization: Math, Stats & CS, Dalhousie University, Halifax, NS, Canada
Summary: Here's what I know
This is in response to Jan Owoc's article about an IBM image processing
setup which is being used by Toronto Police. What I know about stems from
a couple of job interviews I did for the IBM Image Department last summer.
In article <1990May16.184218.28979@calgary.uucp> witold@enme.UCalgary.CA (Witold Jan Owoc) writes:
>
>Today on CTV there was a presentation of such a system used by The
>Toronto Metropolitan Police. It turned out that the job is being done
>by an artist (with a knowledge of a forensic science I presume), as it
>used to be before the computers were invented, only with a computer
>help.
>
>Questions:
>1) Does anybody know what is built into the computer program
> (forensic science knowledge, anatomy of human face, ...) ?
I believe it's strictly image processing and graphical stuff.
e.g. [probably ad-hoc] techniques for doing stuff like removing beards,
plus some super-fancy "paintbrush" type drawing tools. The
system is designed not to replace police artists, but to
make their job faster and easier ... you and I would probably
fail miserably at trying to age a picture :-)
>2) Is it popular setup ? I mean I can order one from IBM ?
No. The project was basically custom built by IBM in collaboration
with the Toronto Police. The Image department at IBM is working on
a number of other major image processing projects as well, with the
the plan of making commercially viable systems eventually (I mean,
we are talking about Big Blue here :-) I suspect that these are mostly
systems requiring more power than you'd have in your home or office,
but there may be PS/2 applications.
>3) What is the technical name of it ?
Dunno.
>4) Is it widely used ? CTV was presenting it as though custom built.
Just by Toronto police, as far as I know.
>5) Anything more about it ? Where (on the net preferably) ?
Probably hard to find out, since it's still in proprietary development
stages at IBM labs in Toronto.
... Sorry I don't know more.
PS I never did get the job (IBM reneged on their commitment to 200 new
positions in the lab). Oh well, I'll just have to go to graduate
school, I guess :-)
Yossarian King | "I plan to live forever ... |
Halifax, Nova Scotia | ... or die in the attempt" |
king@dalcsug.UUCP (if you're lucky) | - [the original] Yossarian |
------------------------------
Date: Fri, 18 May 90 18:09:38 -0400
From: Frederic Leymarie <leyfre@lizard.McRCIM.McGill.EDU>
Subject: TR on Tracking and Shape
I recently completed a Master's thesis in Computer Vision under
Prof. Levine's supervision, with applications to the Cell Tracking
problem. There, I discuss extensively the use of active contour models
such as the "snake", for image segmentation, tracking and shape description
problems.
The following is the abstract and a short Table of Contents:
Title: Tracking and Describing Deformable Objects
Using Active Contour Models
Abstract: In this thesis we consider a number of issues in developing
techniques and algorithms to automate the visual tracking of
deformable objects in the plane. We have applied these techniques
in cell locomotion and tracking studies. We examine two classes
of computer vision problems. First, we consider the segmentation
of a noisy intensity image and the tracking of a nonrigid object.
Second, we consider the shape analysis of an amorphous object.
In evaluating these problems, we explore a new technique based
on an active contour model commonly called a ``snake''. The
snake permits us to simultaneously solve, in constrained cases,
both the segmentation and tracking problems. We present a
detailed analysis of the snake model, emphasizing its limitations
and shortcomings, and propose various improvements to the
original description of the model. Then, we study the two
complementary types of shape descriptors: boundary- and
region-based. We propose to combine these within the context
of the grassfire transform. Two new algorithms are described.
First, we present a contour segmentation technique using
mathematical morphology on the curvature function. Accurate
localization for different scales of curvature features is
achieved. Second, the snake model is used to simulate the
grassfire transform using the previously extracted contour
features. This permits us to produce a multiscale skeleton
representation of shape which is based on the Euclidean
distance metric. New significance criteria for our shape
descriptors, such as the ``region-support'' of curvature
extrema and the ``ridge-support'' of
skeleton branches are also proposed. Finally, numerous
implementation details are discussed; for example, the
description of an optimized sequential Euclidean distance
transform.
Short ToC: Chapter 1 Introduction {1}
Chapter 2 The Snake Model {11}
Chapter 3 Image Segmentation and Cell Tracking {45}
Chapter 4 Shape Features using Curvature Morphology {78}
Chapter 5 Shape Description Using Snakes {105}
Chapter 6 Conclusions {138}
References {143}
Appendix A. Inserting Discontinuities Along the Snake {152}
Appendix B. Imposing Limits on the Snake Forces Amplitudes {158}
Appendix C. Morphological Operations for Functions {161}
Appendix D. Euclidean Distance Mapping in the Discrete Domain {168}
A technical report (TR-CIM-90-9) version of this thesis is available
from our lab:
McGill Research Center for Intelligent Machines,
McGill University, 3480 University St.,
Montreal, Qc., Canada, H3A 2A7.
Frederic Leymarie (PhD student; Computer Vision)
McGill U., E.E. Dept., |tel: (514) 398-7156 (Ext.: 5992)
McConnell Eng. Bldg., Room 424, |FAX: (514) 398-7348
3480 University St., |Email: leyfre!mcgill-vision!uucp
Montreal, Qc., Canada, H3A 2A7. | leyfre@larry.mcrcim.mcgill.edu
------------------------------
Date: 17 May 90 17:29:52 GMT
From: tr@fctunl.rccn.pt (Thomas Rauber)
Subject: bibliographic info
Organization: Universidade Nova de Lisboa -- Lisbon, Portugal
It would be very helpful for my work if I had an easy access to
publications in my working area.
I am working in the context of vision, pattern recognition and
related topics.
Since the bibliographic resources in Portugal are not so well
developed like e.g. in the USA it is difficult to find enough
information.
I would like to know, if there exists a (public, commercial) database
that can be consulted via this net.
I am grateful for any help
Thanks in advance Thomas
Thomas W. Rauber | BITNET/Internet: tr@fctunl.rccn.pt
UNINOVA | UUCP: tr@unl.uucp
Grupo de Robotica Inteligente | Fax: (+351) (1) 295-5641/4461
2825 Monte Caparica, PORTUGAL | Phone: (+351) (1) 295-4464 x.0460
------------------------------
Date: Fri, 18 May 90 07:44:23 EDT
From: flynn@pixel.cps.msu.edu (Patrick J. Flynn)
Subject: Re: Other image processing/vision-related lists.
Status: RO
>Date: Wed, 16 May 90 13:05 N
>From: THIERRY PUN <PUN%CGEUGE51.BITNET@CORNELLC.cit.cornell.edu>
>Subject: Other image processing/vision-related lists.
>
>Computer Vision Bulletin Board Lists
>
>[ several paragraphs deleted... ]
>
>TAAC (90/03/31)
> TAAC board interest group; Pat Flynn (US).
^^^^^^^^^^^^^^
> Broadcast: (US) taac-users@swrinde.nde.swri.edu
> (Europe) taac-users@cs.bris.ac.uk
> Membership: (US) taac-users-request@swrinde.nde.swri.edu
> (Europe) taac-users-request@cs.bris.ac.uk
Keith Pickens at SWRI now administers the US side of the TAAC-Users
list.
Pat Flynn
------------------------------
Date: Fri, 18 May 90 16:04:58 PDT
From: Philip Kahn <vision@deimos.ads.com>
Subject: Updated info on the Vision List and related services
(Especially note the added importance of the Vision List archives.
Please help contribute useful stuff to this archive!)
The INTERNET host site for the Vision List is Vision-List@ADS.COM for
list contributions and Vision-List-Request@ADS.COM for undistributed
correspondence. The Vision List is also distributed as the
COMP.AI.VISION newsgroup. As the moderator, I am interested in
stimulating exchanges on research and other topics of significance to
the vision community. Here is some information which describes the
List and related services.
Requests
--------
If you have problems sending messages to the list, questions about
technical details of the list distribution, or the like, send mail to
Vision-List-Request@ADS.COM and we will respond personally to your
query.
PLEASE DO NOT send your request to be added to/deleted from the list
to Vision-List@ADS.COM. If you do, it may be automatically
redistributed as an article and appear in thousands of mailboxes around
the world. Always send these requests to the -Request address.
Submissions & Release Format
----------------------------
To submit an article to Vision-List, simply send it to:
Vision-List@ADS.COM
Submissions to the list are currently being delivered in a single
batch mailing about once a week and more often as needed. Caution:
the list moderator does not always edit the incoming messages before
they are redistributed. During those weeks, everything operates
automatically, with the attendant advantages and pitfalls. (If you
fall into one of those pits, please send me mail at
Vision-List-Request and we'll do our best to fix things up.)
The following details may help you in formatting and choosing
appropriate titles for your list submissions. Within a single release
of Vision-List, the format consists of an initial identifying header,
a list of topics covered in that release, and then each message
separated by dashed lines (please avoid the use of dashed lines
starting in column 1 in your submissions: they play havoc with most
autodigesters in the networld). Most of the header information is
automatically removed from each submission at release time. When the
software is working correctly, only "Date:", "From:", and "Subject:"
appear.
Archives & Services
-------------------
Backissues of the Vision List Digest and some public domain software
and imagery is available via anonymous FTP. For those of you without
FTP connection, limited backissues may be obtained by mailing your
request to Vision-List-Request@ADS.COM .
To access the Vision List archives from anonymous FTP:
1) FTP to ADS.COM
2) Login name is ANONYMOUS
3) Once you're logged on, change directory (cd) to
pub/VISION-LIST-ARCHIVE
* Backissues of the Vision List Digest may be found in
pub/VISION-LIST-ARCHIVE/BACKISSUES
* Azriel Rosenfeld's Computer Vision bibliographies from 1984-present
may be found in pub/VISION-LIST-ARCHIVE/ROSENFELD-BIBLIOGRAPHIES
* Public domain software (shareware) may be found in
pub/VISION-LIST-ARCHIVE/SHAREWARE
* Imagery may be found in pub/VISION-LIST-ARCHIVE/IMAGERY
** ** ** ** ** **
The list is intended to embrace discussion on a wide range of vision
topics, including physiological theory, computer vision, machine
vision and image processing algorithms, vision techniques to support
robot navigation and spatial representation, artificial intelligence
and neural network techniques applied to vision, industrial
applications, robotic eyes, implemented systems, ideas, profound
thoughts -- anything related to vision and its automation is fair
game.
For those wishing to contribute to the archives with shareware,
imagery, or other good stuff of general interest to the community,
please contact me at Vision-List-Request@ADS.COM .
I hope you enjoy Vision-List!
Phil Kahn, moderator
Vision-List-Request@ADS.COM
------------------------------
End of VISION-LIST
********************
