Copy Link
Add to Bookmark
Report
VISION-LIST Digest Volume 11 Issue 21
VISION-LIST Digest Mon Jun 08 14:20:03 PDT 92 Volume 11 : Issue 21
- Send submissions to Vision-List@ADS.COM
- Vision List Digest available via COMP.AI.VISION newsgroup
- If you don't have access to COMP.AI.VISION, request list
membership to Vision-List-Request@ADS.COM
- Access Vision List Archives via anonymous ftp to FTP.ADS.COM
Today's Topics:
OBVIUS-2.2 image processing package
Clothes Identification
Code for classification & regression trees
Workshop "Autopoiesis & Perception": Call For Participation.
Int. Symposium on 3-D Analysis
ABSTRACT: A general paradigm to generic model-based image understanding
CFV: sci.image.processing
----------------------------------------------------------------------
Date: Fri, 29 May 92 14:05:20 PDT
From: heeger@white.stanford.edu (David Heeger)
Subject: OBVIUS-2.2 image processing package
OBVIUS (Object-Based Vision and Image Understanding System)
OBVIUS is an interactive lisp-based system for manipulating
pictorially displayable objects, called viewables. The most common
type of viewable is a monochrome image, but other examples include
color images, one-bit images, complex images, image pyramids
(including Laplacian pyramids and wavelet pyramids), image sequences,
and discrete functions. Each viewable may be displayed as a picture
on the screen. Each viewable is typically compatible with several
different picture types. For example, a floating point image may be
displayed as an eight bit grayscale picture, as a one bit dithered
picture, or as a graphical surface plot.
The top-level of OBVIUS is implemented in Common Lisp, thus providing
an interpreted, object-oriented programming environment. The
low-level floating point operations are implemented in C for
efficiency. A graphical user interface, based on menus and dialog
boxes is also provided, in addition to the Lisp interpreter
(listener). In the typical mode of interaction, the user types an
expression to the lisp listener (or enters a command in a dialog box)
and it returns a viewable as a result. A picture of that viewable
will then be automatically displayed in a window. Each window
contains a circular stack of pictures. The user can cycle through
this stack using mouse clicks with certain shift (``bucky'') key
combinations. Commonly used operations such as histogram and zoom are
also provided via mouse clicks.
The system provides a library of low-level image processing routines.
Some examples of these are
- arithmetic operations (add, multiply, lookup-table point
operations, etc)
- image statistics (mean, variance, kurtosis, maximum, histograms etc)
- convolutions, Fourier transforms, Hilbert transforms
- geometric operations (crop, slice, rotate, flip-x, etc)
- comparisons (greater-than, etc)
- synthetic image generation
- matrix operations (matrix multiplication, singular value
decomposition, etc)
- data analysis (regression, fitting, statistics)
OBVIUS also provides postscript output of pictures. Writing new
operations in OBVIUS is relatively simple, and it is straightforward
to add new viewable and picture types.
OBVIUS currently runs on Sun workstations with Lucid (Sun) Common Lisp
(version 4.0 or higher), and the X or OpenWindows window system. It
is also recommended that you use the Gnu Emacs text editor with OBVIUS
(also available via anonymous ftp, from prep.ai.mit.edu).
OBVIUS is distributed with complete source code via anonymous ftp
either from white.stanford.edu (IP number 36.121.0.16) or from
whitechapel.media.mit.edu (IP number 18.85.0.125). Documentation
(LaTeX) and a User's Guide with installation instructions are included
in the tarfile. Since it is currently an in-house product it comes
without warrantee or promise of support. Let us know if you ftp a
copy of OBVIUS at your site so that we can send you bug fixes, inform
you of new releases, etc. For more information contact:
David Heeger, heeger@white.stanford.edu, (415) 604-0035
Eero Simoncelli, eero@victoria.media.mit.edu, (617) 253-3891
------------------------------
Date: Wed, 3 Jun 1992 14:00:07 UTC+0200
From: miguel <miguel@iai.es>
Subject: Clothes Identification
Now, we are working on a system to identify a great quantity of
clothes. We have to identify about one millon of clothes in order to
fulfill a contol of them.
In now days, we had thought in two possibilities:
- Bars code, but we would desire the bars code isn't can be
seen. And the clothes are so flexible to scan the bars code in a easy way.
- Invisible Ink. With this procedure we want to simulate the
bars code. We know the invisible inks can be reflect colour lights if we
project ultraviolet light over them. So, we could have a colour code.
Anybody knows other techniques, based on Artificial Vision, to realize
this task? Comments about the last possibilities are wellcome.
Thanks in advance.
Miguel Angel Patricio Guisado.
------------------------------
Date: Mon, 8 Jun 1992 21:07:22 GMT
From: sareibi@cheetah.vlsi.waterloo.edu (Shawki Areibi)
Organization: University of Waterloo
Subject: code for classification & regression trees
hi, I am posting this for a friend.
He is interested in the following alogorithms:
CART "Classification & regression Trees"
AMIG "Average mutual information gain"
If you have any idea where the code for these aglorithms can be found,
I would really appreciate it by sending mail to
sareibi@cheetah.vlsi.waterloo.edu
thanks.
------------------------------
Date: Mon, 8 Jun 1992 14:15 GMT
From: "Barry McMullin, DCU (Dublin, Ireland) <75008378@dcu.ie>" <75008378@dcu.ie>
Subject: Workshop "Autopoiesis & Perception": Call For Participation.
AUTOPOIESIS AND PERCEPTION
A Workshop within ESPRIT BRA 3352
DUBLIN CITY UNIVERSITY: 25-26 August 1992
************ CALL FOR PARTICIPATION ************
A common sense idea of perception is that, through the information
processing capabilities of our sensory/brain system, we come to know
"the" objectively real, external, world. However, this "spectator"
paradigm has not proved very effective (so far) in attempts to build
artificial autonomous systems. It therefore seems appropriate to
critically examine this concept of perception.
One alternative idea is to take a participatory rather than a spectator
view of the relationship between the autonomous system and "the external
world". In this perspective, to perceive is not to process sensory data,
but to apprehend meaning through interaction.
Autopoiesis is an organizational paradigm which can support such a
participatory view of perception. The concept of autopoiesis (lit.
"self-producing"), was introduced to characterise the organisation which
makes living systems autonomous. An autopoietic organisation is one
which is self-renewing (in a suitable environment); autopoietic systems
maintain their organisation through a network of component-producing
processes such that the interacting components generate the same network
of processes which produced them. In the autopoietic paradigm,
perception is an emergent phenomenon characteristic of the interaction
between an autopoietic system and its environment: the system responds
to perturbations in just such a way as to maintain its (autopoietic)
identity.
STRUCTURE:
The key objective of the workshop is to allow for extensive, open,
discussion, and it has been structured accordingly. It will consist of a
small number of prepared papers by invited keynote speakers, punctuated
with extended discussion periods; it will run over one and a half days
(from 9.30 AM on 25th August, to 1.00 PM on 26th August). To
maximize the benefit of the discussion, the workshop will be
limited to 30 participants.
INVITED SPEAKERS (Confirmed):
Prof. Francisco Varela C.R.E.A., Ecole Polytechnique, Paris.
Dr. David Vernon DG XIII, EC Commission, Brussels.
Dr. Dermot Furlong Department of Microelectronic and Electrical
Engineering, Trinity College Dublin.
FURTHER INFORMATION: Barry McMullin, Electronic Engineering,
Dublin City University, Dublin 9, IRELAND.
E-mail: <McMullinB@dcu.ie>
Phone: +353-1-7045432 Fax: +353-1-7045508
************* REGISTRATION FORM *************
The deadline for receipt of registration information is Friday, 31st
July 1992. Due to the limit to 30 participants, early registration is
advisable. However, postal services to Dublin are currently severely
affected by an industrial dispute. Therefore, if you wish to register,
it is recommended that you return this form by E-mail or FAX as soon as
possible, paying the registration fee by Bank Transfer. Please advise if
you require information on hotel accomodation; campus accomodation will
be available at a rate of IRP 20 per night (approx.) - a separate
booking form will be provided on request. The DCU campus is situated in
the north Dublin suburb of Glasnevin, is less than 10 minutes from
Dublin International Airport, and has easy access to the city centre.
All correspondence should be directed to:
Barry McMullin, Electronic Engineering,
Dublin City University, Dublin 9, IRELAND.
E-mail: <McMullinB@dcu.ie> Phone: +353-1-7045432 Fax: +353-1-7045508
Name:...................................................................
Organisation:...........................................................
Address:................................................................
City:........................... Country:..........................
Phone:............... FAX:................. E-mail:...................
Is your organisation a member of the BRA 3352 Working Group on Vision?
YES___ NO___ If YES, which consortium? ...................
Registration Fee: Irish Pounds 60 (or equivalent)
Payment Form: (Check One)
1) Internal Accounting (working group members only) ____
Requires signature of partner representative listed
in BRA 3352 Technical Annex:
Partner Representative:................... Signature................
2) Bank Transfer: ____
Account Name: Dublin City University Conference a/c
Bank: AIB Bank, 7-12 Dame St., Dublin 2, IRELAND.
Account Number: 91765-215 Bank Sorting Code: 93 20 86
(IMPORTANT: Quote your NAME *and* "Ref: 421/01/121 (Autopoiesis)"
in all bank transfer documents.)
3) Bank Draft (made payable to "Dublin City University"): ____
Equivalent of Irish Pounds amount in any EC currency drawn
on a local bank -OR- DM, US$, or Sterling draft drawn on a
UK bank. All charges to be bourn by the remitter.
------------------------------
Date: Thu, 28 May 92 17:32:10 EDT
From: Aissaoui Rachid <aissaoui@ERE.UMONTREAL.CA>
Subject: Int. Symposium on 3-D Analysis
A FIRST ANNOUNCEMENT AND CALL FOR PAPERS
SECOND INTERNATIONAL SYMPOSIUM ON 3-D ANALYSIS OF HUMAN MOVEMENT
Satellite event to the Fourtheen Congress
of the International Society of Biomechanics
Under the patronage of the INTERNATIONAL SOCIETY OF BIOMECHANICS
July 1st - 4th, 1993
Parc du Futuroscope
Poitiers, FRANCE
The International Symposium of 3-D Analysis of Human Movement will be a
scientific and technical forum for investigators of human motion,
whether their work is applied to the study of musculo-skeletal
disability or disease, sport and elite performance, or basic studies
of biomechanics. By generating further communication and contact
between investigators in divers ares, this meeting encourages discussion
to cross the boundaries between scientific disciplines and specialities.
This Symposium will be directed at sharing information and results
relating to philosophies for solving problem on measurement and analysis,
rather than only delivering recent research and study findings.
SCIENTIFIC PROGRAM
The conference is spanned over nearly three days. There will be
five keynote speakers in addition to oral presentations to be paired
with discussion papers. Additional papers will accepted in the form of
poster presentation. A Round Table as well as a Hyde Park Speakers'
Corner are also planned.
1. Themes
Data Capture: New intrumentation and specific hardware, accuracy
and precision of various reconstruction techniques, interpolation and
calibration, lens distortion correction, etc.
Joint Motion: Location of markers, relation between external
reference markers and joint movements, definition of local and global
coordinate systems.
Mechanical Models: 3-D kinematic and kinetic/dynamic joint models,
finite element techniques, estimation of internal forces, model
validation and sensitivity analysis.
3-D Representation: Computer-aided-graphics techniques and animation.
3-D Applications and Interpretation: 3-D variables used in the
analysis of human movement and their relation to convential 2-D
parameters.
2. Submission
Those wishing to submit an abstract are asked to fill out the
Request Form for an author's kit and fax, mail or E-mail it to the
Permanent Secretariat.
TECHNICAL PROGRAM
As a complement to the scientific program, manufacturers of data
capture systems will be given time during the meeting to inform
the participants about their products. There is a possibility for
commercial exhibition if there is a marked interest from the
manufacturers. Those wishing to participate in the Technical
Program are asked to fill out the Request Form and fax, mail or
E-mail it to the Permanent Secretariat.
SOCIAL PROGRAM
There is also a very attractive social program consisting of a
welcoming reception on the evening prior to the meeting (July 1st)
and a tour of the Cognac region to be followed by a gastronomic
dinner in one of the neighboring castles. Throughout the meeting,
the participants will have access to the Parc du Futuroscope site
featuring amazing cinematographic facilities including IMAX, OMNIMAX,
double IMAX, dynamic, 360 degree and 3-D movie theaters to name but
a few!
TRAVEL INFORMATION
Poitiers, a town of 83 000 inhabitants, is about 430 km south-west of
Paris. It is a 90 minute non-stop ride on the TGV (Train a Grande
Vitesse) train, having a cruising speed of 300 km/hr (187 mph)! It can
be taken at the Montparnasse train station in Paris. For those driving,
there is a direct entry from Autoroute A10. There is a number of hotels
on site and a 100 room bloc has been set aside for the Symposium.
REGISTRATION FEE
The fee for the Second International Symposium on 3-D Analysis of Human
Movement is 380 US$ prior to May 1st, 1993 and 450 US$ afterwards. The
registration fees includes meeting material and all all scientific,
commercial and social activities. Registration for the Social Program
only is 90 US$. Checks, money orders or bank drafts drawn in US$ (no
credit cards) are made payable to the International Symposium on 3-D
Analysis of Human Movement and mailed to the Permanent Secretariat.
CANCELLATION POLICY
All cancellations received by June 1st, 1993 will be assessed a 75 US$
non-refundable charge. After this date, there will be NO REFUNDS.
LANGUAGE
The official language of the Symposium is English. Simultaneous
translation will not be available.
DEADLINES
Abstract submission:
FEBRUARY 1st, 1993
Notification to the authors:
APRIL 1st, 1993
Selection of discussors:
APRIL 1st, 1993
Submission of discussors' papers:
MAY 1st, 1993
Early bird registration:
MAY 1st, 1993
ORGANIZATION COMMITTTEE
President Alain JUNQUA, FRA
Scientific Committee
Chairperson Herman J. WOLTRING, NL
Paul ALLARD, CAN
Jacques deGuise, CAN
Melissa GROSS, USA
Zvi LADIN, USA
Bernard LANDJERIT, FR
Chester TYLKOWSKY, USA
Hyde Park Speakers' Corner
Elena BIRYUKOVA, Russia
Round Table
Hans FURNEE, NL
FOR FURTHER INFORMATION AND AUTHOR'S KIT WRITE TO
Dr. Paul ALLARD
Permanent Secretariat
International Symposium on 3-D
Analysis of Human Movement
Centre de recherche
Hopital Sainte-Justine
3175 Cote Ste-Catherine
Montreal, PQ,H3T 1C5
CANADA
tel 1 (514) 345 - 4740
fax 1 (514) 345 - 4801
e-mail AISSAOUI@ERE.UMONTREAL.CA
------------------------------
Date: Fri, 29 May 92 10:25:26 MEZ
From: Pan <ULM101@DBNRHRZ1.rhrz.uni-bonn.de>
Subject: ABSTRACT: A general paradigm to generic model-based image understanding
I'd like to introduce you my recent paper entitled
Production, Inversion and Learning of Spatial Structure:
A general paradigm to generic model-based image understanding
He-Ping Pan (Computer Vision Research Fellow)
Institut fuer Photogrammetrie, Universitaet Bonn
Nussallee 15, 5300 Bonn 1, Germany.
email: ulm101 at dbnrhrz1.bitnet
published in International Archives of Photogrammetry and Remote Sensing
Vol. 29, 1992, Washington. I suppose most of you rarely read the
literature of ISPRS, therefore I give you the abstract below. In case
you are interested, free copy is available on request.
ABSTRACT:
A general paradigm to image understanding is proposed. As knowledge
about the scene captured in a given image plays the central role
to understanding of this image, generic model-based approach aims
at the most compact body of visual knowledge. The dynamics of vision
can be structured in three operations of spatial structure of the
scene: production (genesis) of scene instances from a generic model,
inversion (parsing) of an actual scene instance back to a generic
model, and learning (induction) of a generic model from a set of
provided sample scenes. The plausibility of this general paradigm
will be not only partially proved by theoretical analysis, but
also evidenced by biological facts and psychological empirical
discoveries, as well as supported by research trends in computational
vision. An an instance of this paradigm and an actual application
of this developing theory, the stochastic attributed polygon map
grammars as a generic model of rural landuse maps and remote sensing
images are demonstrated.
CONTENTS:
1 INTRODUCTION
2 SPATIAL STRUCTURE PRODUCTION, INVERSION AND LEARNING SYSTEM (SSPILS)
2.1 Self Model versus World Model
2.2 Three Representation Domains of Scene Structure
2.2.1 The Physical Domain
2.2.2 The Appearance Domain
2.2.3 The Conceptual Domain
2.3 A Taxonomy of Models
2.3.1 Specific Models
2.3.1.1 Maps
2.3.1.2 Shape Fixed Models
2.3.1.3 Number of Parameters Fixed Models
2.3.2 Generic Models
2.3.2.1 Static Descriptive Generalization Models
2.3.2.2 Dynamic Procedural Production Models
2.3.3 Limits of Computationally Based Modeling
2.4 Three Dynamics of Spatial Structure in Vision
2.4.1 Spatial Structure Production
2.4.2 Spatial Structure Inversion
2.4.3 Spatial Structure Learning
2.5 SSPILS Compared with Grammar
2.6 Explicit Computer Vision --- A foundation of Computer Vision
as a discipline of science
3 STOCHASTIC ATTRIBUTED POLYGON MAP GRAMMARS: A CASE STUDY
3.1 Polyplex versus Simplex-Complex
3.1.1 Polyplex
3.1.2 Simplex-Complex
3.2 Formalization of Grammar
3.2.1 Grammar
3.2.2 Primitives and Relations
3.2.3 Productions
3.3 Parsing of Segmented Images (Inversion)
3.3.1 Building the Polygon Split Tree bottom-up
3.3.2 Error-Correcting Parsing of Segmented Images
4 CONCLUSIONS
REFERENCES
------------------------------
Date: Thu, 4 Jun 92 15:46:50 -0400
From: John Stanley <stanley@oce.orst.edu>
Subject: CFV: sci.image.processing
Organization: oce.orst.edu
Followup-To: poster
[ As I have noted earlier, I do not advocate you vote against this: its
charter and subject matter is contained within the scope of the
Vision List (comp.ai.vision). It will serve to divide our readership.
To vote against this group, mail to mail-server@pit-manager.mit.edu
with the subject "vote sci.image.processing no".
phil... ]
NAME: sci.image.processing
STATUS: Unmoderated
sci.image.processing Scientific image processing and analysis
CHARTER:
sci.image.processing provides a forum for discussion of the scientific
uses of image processing and analysis. Discussions of algorithms
and application programs are appropriate, with emphasis on solving real
world image processing problems. Discussions of computer graphics,
visualization, output, window systems, communications, etc. are not
appropriate. Posting of images is strongly discouraged. Images should
be posted to alt.binaries.pictures.misc, or made available for FTP.
This group will take a broad definition of the term image processing.
Most topics related to the formation and analysis of images
are appropriate so long as a more focused newsgroup does not exist.
This presently excludes topics such as computer vision (comp.ai.vision)
and image compression per se (comp.compression), but not topics such
as image detectors or calibration of imaging equipment. Obviously,
topics such as the impact of lossy image compression on image processing
algorithms or the use of AI to analyze scientific images are not excluded.
One purpose is to develop and maintain Frequently Asked Question (FAQ)
articles on the subject. Another is to provide a forum where
experimentalists learn about image processing algorithms while computer
scientists hear about the uses to which generic algorithms are put, or
the requirements of a specific application. Both will discuss what can
and cannot be done to improve image quality before and during processing.
The "IEEE Transactions on Image Processing" gives a list of topics
which they consider appropriate for that journal. Some of those
topics are also appropriate here:
Image Processing
Coding Filtering Enhancement
Restoration Segmentation
Multiresolution Processing Multispectral Processing
Image Representation Image Analysis
Interpolation and Spatial Transformations
Motion Detection and Estimation Image Sequence Processing
Video Signal Processing Noise Modeling
Architectures and Software
Computed Imaging
Acoustic Imaging Radar Imaging Tomography
Magnetic Resonance Imaging Geophysical and Seismic Imaging
Radio Astronomy Speckle Imaging
Computer Holography Confocal Microscopy
Electron Microscopy X-ray Crystallography
Coded-Aperture Imaging Real-Aperture Arrays
RATIONALE:
While some existing newsgroups carry traffic related to image processing,
there are many aspects to scientific imaging and no newsgroup addresses
more than a few of them. Image processing is usually carried out by
software, but many questions appropriate to this group have little or
nothing to do with software.
By placing the group in the sci.* hierarchy, it should be clear to
new users that the the group is not limited to software or algorithms.
This placement also allows for the future discussion of a sci.image.*
hierarchy with minimal upheaval.
An unmoderated group is needed to permit rapid responses to queries.
A focused, medium noise group is needed in order to attract and retain
the readership which is essential to obtaining knowledgeable answers.
VOTING INSTRUCTIONS:
Do not post your vote, it will not be counted. The reply-to address of
this posting should point to the correct place, but double check your
mail before sending to make sure.
To cast a vote, send mail to
"mail-server@pit-manager.mit.edu"
with ONE of the following strings (you don't need the quotes)
"vote sci.image.processing yes"
"vote sci.image.processing no"
on a line by itself in the subject field or body of the message, where
"yes" means that you DO want sci.image.processing created.
"no" means that you DO NOT want sci.image.processing created.
* The deadline is 11:59:59 PM Tue 30-Jun-1992, EDT (GMT - 4).
* The mail server will reply within two hours confirming your vote.
If no confirmation is received (allowing for normal E-mail round
trip time), you may inquire at postmaster@pit-manager.mit.edu to
verify that the vote was received.
* If your mail message contains a signature, then be sure to put the
string "quit" before it to prevent the mail server from being confused.
* DO NOT mail votes to John Stanley <stanley@oce.orst.edu>
* DO NOT mail votes to jik@mit.edu.
* DO NOT include any message in your vote (it won't be read).
* Offically, the vote is being run by John Stanley <stanley@oce.orst.edu>
However, the collection and automation of the vote taking process is
kindly being performed by Jonathan Kamens (jik@mit.edu).
software, this CFV is being posted three times. The first posting will
be to the groups shown in the header of this posting, which will include
news.announce.newgroups and news.groups.
As soon as the CFV appears in news.announce.newsgroups, it will be posted
sci.astro, sci.bio, sci.math, sci.med, sci.research,
sci.space, sci.geo.geology, sci.geo.meteorology,
sci.geo.fluids, comp.infosystems.gis, bionet.general,
comp.graphics.avs, comp.soft-sys.khoros
In addition, the CFV will be posted indivudually to the following moderated
bionet.announce biosci-announce-moderator@genbank.bio.net
bionet.biology.computational comp-bio-moderator@genbank.bio.net
comp.ai.vision vision-list@ads.com
comp.graphics.research graphics@scri1.scri.fsu.edu
This division is solely to bypass the limitations of line length in some
versions of news software.
------------------------------
End of VISION-LIST digest 11.21
************************
