Copy Link
Add to Bookmark
Report
VISION-LIST Digest Volume 12 Issue 20
VISION-LIST Digest Wed Apr 28 14:01:29 PDT 93 Volume 12 : Issue 20
- ***** The Vision List has changed hosts to TELEOS.COM *****
- Send submissions to Vision-List@TELEOS.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@TELEOS.COM
- Access Vision List Archives via anonymous ftp to FTP.TELEOS.COM
Today's Topics:
Looking for the origin of two formuli
Pan/Tilt Mechanisms
RasterOps VideoLive Card
Announcement/Call for IS&T/SPIE Image and Video Storage and Retrieval
Final call for papers
CVPR'93 Tutorial Description
Final Program for Conference on Understanding Images
----------------------------------------------------------------------
Date: Wed, 28 Apr 1993 10:29:27 +0200
From: Chokri BENAMAR <Chokri.Benamar@lai1.univ-lyon1.fr>
Subject: looking for the origin of two formuli
I just finish reading the following paper " Cowan,C.K., and
Kovesi,P.D. Automatic sensor placement from vision task requirements.
IEEE Transactions on Pattern Analysis and Machine Intelligence 10,3
(May 1988), 407-416". This paper deals with the problem of
automatically generating the possible camera locations for observing
an object.The approach uses some geometrical constraints as
resolution,focus,field of view,visibility,etc... that should be
satisfied to generate the three dimensional region of camera
viewpoints. For the focus constraint, the region of viewpoints is
determined such all points on the surface(s) to be observed are in
focus.For any camera and lens setting, there is only one distance,
called the focus distance , at wich a point is perfectly focused on
the image plane.The maximum distance D1 and the minimum distance D2 of
points in focus are related to the focus distance D, the lens's focal
length f,the diameter of the lens's aperture a, and minimum dimension
of a camera pixel c as follow: D1=Daf/(af-c(D-f)) and
D2=Daf/(af+c(D-f)); My question is what's the origin of these last
formuli and how are they demonstrated. Thanks for help.
Chokri BEN AMAR
LAI,BAT.303,INSA,20 Avenue A. Einstein
69621 Villeurbanne FRANCE
TEL:(33)72.43.81.98
FAX:(33)72.43.85.35
E-MAIL:benamar@lai1.univ-lyon1.fr
------------------------------
Date: Wed, 28 Apr 93 8:03:06 EDT
From: "William K. Cadwallender" (FSAC-ND) <cadwall@PICA.ARMY.MIL>
Subject: Pan/Tilt Mechanisms
We are looking for a fast, rugged computer controller pan/tilt camera mount for
a mobile platform. What are robotics groups out there using for camera mounts?
Please let us know what you are using, and we will summarize the results. If we
can find something close, we can possibly provide some $$$ to modify an existing
product. Please respond directly to jlamie@PICA.ARMY.MIL> (He's the M.E> tasked
with coming up with a pan/tilt, his name is Jan Lamie).
William Cadwallender <cadwall@PICA.ARMY.MIL>
Mobile Detection, Assessment,Response System (MDARS) Lab
------------------------------
Date: Wed, 28 Apr 1993 07:41:08 GMT
From: paulus@immd5.informatik.uni-erlangen.de (Dietrich Paulus)
Organization: Regionales Rechenzentrum Erlangen, Germany
Subject: RasterOps VideoLive Card
Summary: Software needed for RasterOps VideoLive Card on HP (hpux 9.x)
Keywords: software request
We are using a RasterOps VideoLive Card for frame grabbing on an HP 9000/735.
The X-Interface is nice for demonstrations. However, for real image processing
applications we need to access the card directly. As an example, we want to
capture pictures continuously with the maximum possible rate and shuffle the
data to main memory.
All requests to HP failed (up to now). Does anyone have software or experience
for this kind of problems? Any answers or hints are appreciated!
Thanks in advance!
Dr. Dietrich Paulus Universitaet Erlangen-Nuernberg
Informatik 5, Mustererkennung (Prof. Dr. H. Niemann)
Martensstr. 3 Tel. 09131-857894 / 857774
D-8520 Erlangen FAX: 09131-303811
FRG paulus@informatik.uni-erlangen.de
------------------------------
Date: Wed, 28 Apr 93 12:25:41 PDT
From: "Wayne Niblack" <niblack@almaden.ibm.com>
Subject: Announcement/Call for IS&T/SPIE Image and Video Storage and Retrieval
Announcement and Call for Papers
SPIE / IS&T Symposium on Electronic Imaging Science and Technology
Storage and Retrieval for Image and Video Databases II
February 6 - 10, 1994
San Jose Convention Center
San Jose, California
Conference Chairs
Wayne Niblack
IBM Almaden Research Center
Dr. Ramesh Jain
University of California, San Diego
Storage and retrieval for image and video databases
Vast on-line collections of image, video, and multimedia data are
becoming feasible as a result of advances in technology. These
collections have a wide range of applications in medicine, remote
sensing, industry, engineering, digital video production, education,
entertainment, marketing, and on-line information services. One of
the challenges in using this data is how to model, organize, and index
it for efficient storage and retrieval.
This conference will address issues in storage organization and
hierarchies, delivery of real-time data, "intelligent" content-based
retrieval, automatic and semi-automatic methods of image annotation,
definition and computation of features for image and video indexing,
and modelling of image and video data for storage and retrieval.
Papers are solicited in the following areas:
* Image and video databases
* Storage management for large image/video databases
* Intelligent retrieval methods such as content-based
indexing and geometric/structural hashing
* Query by image example and/or sketch
* Indexing by color, texture, shape, layout, and motion
* Visual query languages and user interface for query
* Semantic queries based on image descriptors
* Similarity retrieval for image and video data
* Image and/or digital video applications
* Neural network, feature-based, and information theoretic
approaches to image and video storage, indexing, and retrieval
* Applications and systems of image and video databases in areas
such as biomedical, publishing, journalism, GIS, and cataloging
Submit a 500 word abstract by July 12, 1993 to abstract@mom.spie.org
(ASCII format), via fax to SPIE at 206-647-1445 or via mail to
IS&T/SPIE, PO Box 10, Bellingham, WA 98227-0010, telephone 206 676 3290.
Include in your submission:
1. Conference title and chairperson
2. Authors full names, complete address (including telephone, fax,
mail, and e-mail), and brief author biography (50 to 100 words,
principal author only).
3. Paper abstract
Abstract due date: July 12, 1993
Acceptance notification: October 21, 1993
Camera Ready Abstract due date: November 15, 1993
Manuscript due date: January 10, 1994
------------------------------
Date: 28 Apr 1993 20:15:19 GMT
From: sull@focus.csl.uiuc.edu (Sanghoon Sull)
Organization: Center for Reliable and High-Performance Computing, University of Illinois at Urbana-Champaign
Subject: Final call for papers
Summary: Final CFP: Neural networks on shape analysis
Keywords: paper,neural networks,shape analysis
CALL FOR PAPERS
Progress In Neural Networks
Special Volume on Shape Analysis
Omid M. Omidvar Minsoo Suk
Series Editor Volume Editor
Significant progress has been made recently in shape analysis using
neural networks and the energy minimization concept. Ablex Publishing
Corporation is planning a special volume on "Shape Analysis", scheduled
for this year.
This volume will be a part of "Progress in Neural Networks," an annual
book series reviewing research in modelling, analysis, design and
application of neural networks. The primary aim of this volume is to
present, in a single volume, the most important achievements made on
this important topic, which are otherwise scattered in diverse literature.
Authors are invited to submit original manuscripts detailing recent
progress. Suggested topics include, but are not limited to: shape modelling,
shape estimation, shape recovery, shape representation, shape matching,
surface reconstruction and surface decomposition. Research work utilizing
neural networks directly or based on energy minimization techniques such as
Markov random field, mean field annealing, simulated annealing, graduated
non-convexity algorithms and resistive networks are most welcome.
The paper should be tutorial in nature, self contained and preferably,
but not necessarily, about fifty double spaced pages in length. Please send
electronically (if you prefer, you can send a hardcopy to the address below)
an abstract and an outline to
msuk@ima.enst.fr
by May 30, 1993. The full paper must be submitted by July 31, 1993 to:
Europe USA
Professor Minsoo Suk Professor Omid M. Omidvar
Department IMAGES Computer Science Department
Ecole Nationale Superieure University of District of Columbia
des Telecommunications 4200 Connecticut Ave. N.W.
46, rue Barrault Washington D.C. 20008
75634 PARIS CEDEX 13 Phone: (202)282-7345
FRANCE Fax: (202)282-3677
Fax:33-1-45-81-37-94 email: oomidvar@udcvax.bitnet
email:msuk@ima.enst.fr
------------------------------
Date: Wed, 28 Apr 1993 16:36:11 -0400
From: taubin@watson.ibm.com (Gabriel Taubin)
Subject: CVPR'93 Tutorial Description
The following tutorial is being offered at the CVPR '93 conference.
Information about the other tutorials can be found in a previous issue
of the Vision-List (12.16). Any tutorial with insufficient advance
registration is subject to cancellation (with full refund to those who
did register).
Specific questions regarding this tutorial may be directed to the
presenters. General questions about the CVPR tutorial program can be
directed to Dr. Terry Boult <tboult@cs.columbia.edu>. The CVPR
advance program, registration form, and other helpful information can
be ftp'ed from cs.columbia.edu:/pub/vision/CVPR93 .
CVPR'93 TUTORIAL DESCRIPTION
Tutorial #4, Monday 6/14/93 8:00, am - 12:00 pm
OBJECT RECOGNITION WITH ALGEBRAIC GEOMETRY,
INVARIANTS, AND BAYESIAN METHODS
A Powerful New Technology For 2D And 3D Object Recognition.
David B. Cooper Gabriel Taubin
Brown University IBM T.J. Watson Research Center
cooper@lems.brown.edu taubin@watson.ibm.com
This tutorial will show how concepts from algebraic geometry, invariant
theory and probability theory can be integrated in a computationally
efficient framework for object recognition and positioning. This new
technology is ideally suited to handling complex free-form objects in
addition to handling restricted shapes such as conics. The following
topics, which are of use individually in addition to being presented as
an integrated technology, will be covered :
1) ALGEBRAIC CURVE AND SURFACE MODELING:
Algebraic curves and surfaces, defined implicitly as sets of zeros of
polynomials, and generalizing the case of quadrics to higher degree,
have remarkable modeling power. Single 3rd or 4th degree curves can
represent all the hand printed lower case characters. Algebraic curves
and surfaces can represent complex subparts of airplanes or military
vehicles, they can represent free-form solids such as fruits, and
curved manufactured parts. Unlike Fourier series, they can represent
curves and surfaces which intersect themselves and consist of
disconnected components. They can approximate superquadrics and much
more complicated shapes. The exact same formulation handles 2D curves
and 3D surfaces, and with simple extension also nonplanar curves in 3D.
2) ALGEBRAIC CURVE AND SURFACE FITTING:
Algebraic curves and surfaces are fitted to unorganized sets of data
points using nonlinear least-squares methods. Ideally, the sum of the
squares of the Euclidean distances from the data points to the curve or
surface should be minimized, but this is impractical because in general
there is no closed-form expression for the Euclidean distance from a
point to an algebraic curve or surface. Approximate distances are used
instead. We will describe in detail a number of approximations that
make these fitting algorithms computationaly efficient and robust.
Fitting can be done in real time on a SPARC station. The zero sets of
polynomials may be unrully. Though a zero set may fit a data set very
well, it may also have unwanted parts elsewhere. We will present a
solution to this problem.
3) INVARIANT CLASSIFICATION:
Object recognition can be realised by comparing the coefficients of the
polynomial which defines the best fitting algebraic curve or surface to
the data with stored coefficient vectors for polynomials in a data
base, one polynomial for each stored object. Unfortunately, these
coefficients are functions of the orientation, location, and possible
linear transformation of the object sensed in the data. Fortunately,
there are functions of the coefficients of the polynomials, called
Euclidean invariants, that are invariant to data position and
orientation, and affine invariants that are invariant to data position
and general linear transformations. These invariants are functions
only of the object shape and can be used for object recognition. We
will describe a systematic method to generate expressions for these
invariants and to evaluate them efficiently using matrix computations.
4) POSITIONING:
Very often, once an object is classified as being an instance of a
particular model from the database because the corresponding invariants
match, it is necessary to compute the matching transformation. With
the same tools and concepts used to compute invariants, we will
describe how to define and compute the center and principal axes of an
algebraic curve or surface of arbitrary degree, generalizing the well
known case of the ellipse. The matching transformation between two
curves or surfaces of the same degree will be obtained by matching the
corresponding centers and principal axes.
4) BAYESIAN METHODS:
Unfortunately, sometimes the coefficients and the invariants of
polynomials fitted to data are highly sensitive to noise and to missing
data--the latter being prevalent in the presence of partial occlusion
or higher degree. Fortunately, this variability is overcome extremely
effectively through the use of asymptotic Bayesian methods. We will
show how to express the probability of a data set for use with a
parameterized 2D curve or 3D surface model, how to determine the
appropriate degree of the implicit polynomial representation, derive
simply computable Bayesian recognizers for individual objects or
classes of objects, and show how to use the technology for modeling and
recognizing complex objects represented by a number of implicit
polynomial patches. Lastly, we will show how to move a sensor in order
to collect data so as to recognize an object as quickly as possible.
------------------------------
Date: Tue, 27 Apr 93 22:06:20 EDT
From: "Dr. Francis T. Marchese" <MARCHESF@PACEVM.bitnet>
Subject: Final Program for Conference on Understanding Images
FINAL PROGRAM ANNOUNCEMENT FOR THE CONFERENCE ON
UNDERSTANDING IMAGES
SPONSORED BY THE ACM/NYC SIGGRAPH AND
PACE UNIVERSITY'S SCHOOL OF COMPUTER SCIENCE AND INFORMATION SYSTEMS
ON
FRIDAY & SATURDAY, MAY 21-22, 1993
AT
THE PACE DOWNTOWN THEATER, 1 PACE PLAZA, NEW YORK, NY 10038
IN HYPERMDIA, MULTIMEDIA AND VIRTUAL REALITY SYSTEMS VAST AMOUNTS OF
INFORMATION CONFRONT THE OBSERVER OR PARTICIPANT. YET, IMAGE CONSTRUCTIO
, TRANSMISSION, RECEPTION, DECIPHERMENT AND ULTIMATE UNDERSTANDING ARE
COMPLEX TASKS STRONGLY INFLUENCED BY PHYSIOLOGY, EDUCATION AND CULTURE.
THEREFORE, THE PURPOSE OF THIS CONFERENCE IS TO BRING TOGETHER A BREADTH
OF DISPCIPLINES, INCLUDING PHYSICAL, BIOLOGICAL, COMPUTATIONAL SCIENCES,
TECHNOLOGY, ART, PSYCHOLOGY, PHILOSOPHY, AND EDUCATION, IN ORDER TO
DEFINE AND DISCUSS THE ISSUES ESSENTIAL TO IMAGE UNDERSTANDING WITHIN TH
COMPUTER GRAPHICS CONTEXT.
PROGRAM SCHEDULE:
8:00AM-8:30AM REGISTRATION
8:30AM-9:00AM CONTINENTAL BREAKFAST
9:00AM-9:15AM OPENING ADDRESS - DR. SUSAN M. MERRITT, DEAN, SCHOOL OF
COMPUTER SCIENCE AND INFORMATION SYSTEMS, PACE UNIVERSIT
9:15AM-10:00AM "DESIGNING TECHNOLOGY: A CHALLENGE FOR ALL DESIGNERS"
10:00AM-10:30AM "PHOTOGRAPHIC INTERPRETATION"
TOM HUBBARD, SCHOOL OF JOURNALISM, OHIO STATE UNIVERSITY
10:30AM-11:00AM *BREAK*
11:00AM-11:45AM "COMPOSING AND UNDERSTANDING SPATIAL IMAGES"
LES M. SZTANDERA, ELECTRICAL ENGINEERING DEPARTMENT,
UNIVERSITY OF TOLEDO
11:45AM-12:30PM "SOME SPECULATIONS ABOUT GRAPHIC COMMUNICATION"
BARBARA TVERSKY, PSYCHOLOGY DEPARTMENT, STANFORD UNIV.
12:30PM-2:00PM *LUNCH BUFFET*
2:00PM-2:45PM "AUTOMATING PROCEDURES FOR GENERATING CHINESE CHARACTERS
JOHN LOUSTAU AND JONG-DING WANG, MATH AND STAT DEPARTMEN
HUNTER COLLEGE
2:45PM-3:30PM "IMPLEMENTATION OF COLLABORATIVE MULTIMEDIA TECHNOLOGIES
IN URBAN PLANNING SITUATIONS"
MICHAEL J. SHIFFER, COMPUTER RESOURCE LABORATORIES, MIT
3:30PM-4:00PM *BREAK*
4:00PM-4:45PM "GESTURE TRANSLATION: USING CONVENTIONAL MUSICAL
INSTRUMENTS IN UNCONVENTIONAL WAYS"
ROBERT WILLIAMS, COMPUTER SCIENCE DEPARTMENT, PACE UNIV.
4:45PM-5:30PM "THE RULING EFFECT OF CONTOURS, SURFACE MARKINGS AND
BACKGROUND IN PERCEPTION OF SHAPE FROM SHADING"
XIAOPING HU AND NARENDA AHUJA, BECKMANN INSTITUTE AND
DEPARTMENT OF ECE, UNIVERSITY OF ILLINOIS
SATURDAY, MAY 22ND, 1993
8:30AM-9:00AM CONTINTENTAL BREAKFAST
9:00AM-9:45AM "VISUALIZATION FOR THE DOCUMENT SPACE"
XIA LIN, LAW LIBRARY, PACE UNIVERSITY
9:45AM-10:30AM "VISUAL LANGUAGE"
JUDSON ROSEBUSH, JUDSON ROSEBUSH COMPANY, NYC
10:30AM-11:00AM *BREAK*
11:00AM-11:45AM "NEUROMUSIC"
MATTHEW WITTEN AND ROBERT WYATT, UNIVERSITY OF TEXAS
AND CENTER FOR HIGH PERFORMANCE COMPUTING
11:45AM-12:30PM "MASACCIO'S BAG OF TRICKS"
MARC DE MEY, UNIVERSITY OF GHENT
12:30PM-2:00PM *LUNCH BUFFET*
2:00PM-2:45PM "IS ALLIGATOR SKIN MORE WRINKLED THAN TREE BARK? THE
ROLE FO TEXTURE IN OBJECT DESCRIPTION"
A. RAVISHANKAR RAO, IBM WATSON RESEARCH CENTER AND
NALINI BHUSHAN, PHILOSOPHY DEPARTMENT, SMITH COLLEGE
2:45PM-3:30PM "UNIVERSALITY AND VARIABILITY IN HUMAN VISUAL INFORMATIO
PROCESSING"
BEVERELY J. JONES, SCHOOL OF ARCHITECTURE AND APPLIED
ARTS, UNIVERSITY OF OREGON
3:30PM-4:00PM *BREAK*
4:00PM-4:45PM "AESTHETICS AND NATURE: THE MANUFACTURING OF AN
AUTHORITATIVE VOICE IN SCIENTIFIC VISUALIZATION"
MARK BAJUK, NCSA, UNIVERSITY OF ILLINOIS
4:45PM-5:30PM "THE METAPHYSICS OF VIRTUAL REALITY"
5:30PM-6:15PM "SONIC ISSUES"
RORY STUART, NYNEX
HOTEL ACCOMMODATIONS:
THE HOTEL MILLENIUM, 55 CHURCH STREET, NEW YORK, NY 10048
VOICE: (212) 693-2001 (800) 835-2220 ; FAX (212) 571-2317
PLEASE CALL HOTEL DIRECTLY FOR RESERVATIONS AND MENTION PACE UNIVERSITY/
NYC/ACM SIGGRAPH. RESERVATIONS MUST BE MADE BY MAY 10, 1993.
SPECIAL CONFERENCE RATE: $125 PER NIGHT SINGLE/DOUBLE
CHECK-OUT: 2PM CHECK-IN: 3PM (NOTE: HOTEL IS WITHIN WALKING DISTANCE)
AIR TRAVEL ACCOMODATIONS:
CONTINENTAL AIRLINES 800-468-7022 OR CLASSIC TRAVEL & TOURS AGENCY
800-875-2765; PLEASE CALL DIRECTLY FOR RESERVATIONS & USE #ZEB82.
CONFERENCE ON UNDERSTANDING IMAGES - REGISTRATION FORM
NAME: ____________________________
TITLE: ____________________________
COMPANY: ____________________________
ADDRESS: ____________________________
CITY: ____________________________ STATE:_____ ZIP:__________
DAY PHONE: ____________________________ EVENING PHONE: ___________
EMAIL: ____________________________ FAX: _____________________
REGISTRATION FEES:
(INCLUDES CONTINENTAL BREAKFAST, BREAKS AND LUNCH)
PRE-REGISTRATION (POSTMARKED BY MAY 1, 1993)
___ ACM/SIGGRAPH MEMBER $55 ___ NON-MEMBER $75
___ STUDENT REGISTRATION $40
REGISTRATION ____ ON-SITE OR AFTER MAY 1, 1993 $95
PLEASE MAKE CHECKS PAYABLE TO NYC/ACM SIGGRAPH AND REMIT BY MAY 1, 1993
SEND REGISTRATION FEES AND INFORMATION TO:
DR. FRANCIS T. MARCHESE
COMPUTER SCIENCE DEPARTMENT
NYC/ACM SIGGRAPH CONFERENCE
1 PACE PLAZA ROOM T-1704
NEW YORK, NY 10038
VOICE: 212-346-1803 FAX: 212-346-1933
EMAIL: MARCHESF@PACEVM.BITNET
------------------------------
End of VISION-LIST digest 12.20
************************
