Copy Link
Add to Bookmark
Report
VISION-LIST Digest 1988 08 01
Vision-List Digest Mon Aug 1 12:31:07 PDT 1988
- Send submissions to Vision-List@ADS.COM
- Send requests for list membership to Vision-List-Request@ADS.COM
Today's Topics:
-- Thesis defense: Bayesian Modeling of Uncertainty in Low-Level Vision
-- Automated supernova search people need advice
-- Workstation Questions
-- NETWORK COMPUTING FORUM - CALL FOR PAPERS
-- FIRST ANNUAL MEETING OF THE INTERNATIONAL NEURAL NETWORK SOCIETY
-- Public Domain Sun Image Processing Software
-- multidisciplinary conference on VISION, PROCESSING, and DISPLAY
----------------------------------------------------------------------
Date: Mon, 18 Jul 1988 09:34-EDT
From: Richard.Szeliski@IUS2.CS.CMU.EDU
Subject: Thesis defense: Bayesian Modeling of Uncertainty in Low-Level Vision
Bayesian Modeling of Uncertainty in Low-Level Vision
[ Thesis defense ]
Richard Szeliski
Computer Science Department
Carnegie Mellon University
July 28, 1:00pm, WeH 5409
ABSTRACT
Over the last decade, many low-level vision algorithms have been devised for
extracting depth from one or more intensity images. The output of such
algorithms usually contains no indication of the uncertainty associated with
the scene reconstruction. In other areas of computer vision and robotics,
the need for such error modeling is becoming recognized, both because of the
uncertainty inherent in sensing, and because of the desire to integrate
information from different sensors or viewpoints.
In this thesis, we develop a new Bayesian model for the dense fields such as
depth maps or optic flow maps that are commonly used in low-level vision.
The Bayesian model consists of three components: a prior model, a sensor
model, and a posterior model. The prior model captures any a priori
information about the structure of the dense field. We construct this model
by using the smoothness constraints from regularization to define a Markov
Random Field. By applying Fourier analysis to this prior model, we show
that the class of functions being modeled is fractal. The sensor model
describes the behaviour and noise characteristics of our measurement system.
We develop a number of sensor models for both sparse depth measurements and
dense flow or intensity measurements. The posterior model combines the
information from the prior and sensor models using Bayes' Rule, and can be
used as the input to later stages of processing. We show how to compute
optimal estimates from the posterior model, and also how to compute the
uncertainty (variance) in these estimates.
This thesis applies Bayesian modeling to a number of low-level vision
problems. The main application is the on-line extraction of depth from
motion. For this application, we use a two-dimensional generalization of
the Kalman filter to convert the current posterior model into a prior model
for the next estimate. The resulting incremental algorithm provides a dense
on-line estimate of depth whose uncertainty and error are reduced over time.
In other applications of Bayesian modeling, we use the Bayesian
interpretation of regularization to choose the optimal smoothing parameter
for interpolation; we use a Bayesian model to determine observer motion from
sparse depth measurements without correspondence; and we use the fractal
nature of the prior model to construct multiresolution relative surface
representations. The approach to uncertainty modeling which we develop, and
the utility of this approach in various applications, support our thesis
that Bayesian modeling is a useful and practical framework for low-level
vision.
------------------------------
Date: Fri, 29 Jul 88 07:46:22 PDT
From: John B. Nagle <jbn@glacier.stanford.edu>
Subject: Automated supernova search people need advice
This is forwarded from USENET; please reply to "beard@uxl.lbl.gov", not me.
>From: beard@ux1.lbl.gov (Patrick C Beard)
Subject: Workstation Questions
Date: 28 Jul 88 23:41:21 GMT
Distribution: comp.sys.workstations,comp.edu,comp.graphics,comp.os.vms
Organization: Lawrence Berkeley Laboratory, Berkeley
Summary: Questions about available workstations, accelerating mVax
Hello everybody.
My group is conducting an automated search for supernovae and is
in the market for upgrading the computer that is the heart of the
system. We require a system that outperforms our current computer,
a Microvax, by at least a factor of 4, and hopefully a factor of 10.
I am submitting this message to the network community to ask three
questions:
1) What machines are there in the workstation type class that can
outperform a Microvax by a factor of 4 to 10 times? (Please describe
briefly, cost, speed, manufacturer.)
2) Alternatively, what options exist for speeding up a Microvax?
Are there accelerator boards, processor upgrades, anything you can
think of?
3) What image processing systems are available? Commercial or public
domain, source code included, optimized for what hardware, and how easy
are they to modify or extend for special purpose use (such as astronomical
work)?
You may answer me directly, or via the net. I'm sure there are a lot
of people who could benefit from this information.
Thanks in advance,
+=============================================================+
| Patrick C. Beard |
| Lawrence Berkeley Laboratory |
| Automated Supernova Search |
+-------------------------------------------------------------+
| PCBeard@LBL.gov (arpa only) |
+=============================================================+
------------------------------
Date: 29 Jul 88 12:31 PDT
From: William Daul / McAir / McDonnell-Douglas Corp <WBD.MDC@OFFICE-8.ARPA>
Author: Beverly Pieper <BKP.MDC@office-8.arpa>
Subject: NETWORK COMPUTING FORUM - CALL FOR PAPERS
NETWORK COMPUTING FORUM
CALL FOR PAPERS
OCTOBER 5-8, 1988
HOLIDAY INN WESTPORT, ST. LOUIS, MISSOURI
The next meeting of the Network Computing Forum will be held on October 5-7 in
St. Louis, Missouri. This will be the fourth meeting of the Forum, and will
focus on the role of the Forum as a catalyst for change in the industry. The
Forum is an industry group chartered to lead the way for rapid adoption of
multi-vendor network computing concepts and technologies. Forum meetings allow
representatives from users and vendors to work together on common issues in an
open, informal atmosphere. The Forum has over 100 member organizations, and
more than 220 representatives attended the May 1988 meeting.
Forum meetings are organized into three sessions: a conference featuring
invited papers and panel sessions, meetings of interest groups and working
groups, and a policy making executive committee meeting. Some areas of
interest to the Forum member organizations are listed, to suggest possible
topics for papers:
Definition of user requirements for network computing
Practical experiences using network computing concepts & technologies
Partitioning and/or integration of applications across networks
Remote procedure calls and other core services for network computing
System and network administration for networks of heterogeneous computers
User interfaces and user environments for network computing
Software licensing in a network environment
Data representation and command scripting across heterogeneous networks
Use of network computing with IBM mainframes (MVS and VM)
Invited Papers
As part of each Forum meeting, papers are invited from the community at
large for presentation and discussion. These papers should address the use
or development of network based applications and services. Emphasis should
be placed on creating and using tightly coupled links between multiple,
heterogeneous computer systems. Technical descriptions of research
projects, user experiences, as well as commerically available products are
welcome. Invitations are also extended for more informal talks on practical
experience in administering heterogeneous computer networks. All
presentations should be 35 minutes in length, with 15 minutes of discussion
following each presentation.
Abstracts must be received by August 10, 1988. Abstracts should summarize
the paper in two or three paragraphs and include the mailing address,
affiliation, and phone number of the author(s). Notification of abstracts
selected will be sent on August 19, 1988 and papers must be submitted no
later than September 20, 1988. Papers can be copyrighted, but must include
authorization for unrestricted reproduction by the Network Computing Forum.
Papers can be marked as working papers to allow future publication.
SEND ABSTRACTS BY AUGUST 10, 1988 TO the Program Chairman for the October 1988
meeting:
T.D. Carter
c/o Jan McPherson
McDonnell Douglas Travel Company
944 Anglum Drive, Suite A
Hazelwood, MO 63042
(314) 233-2951
Internet Address: TDC.MDC@OFFICE-8.ARPA
------------------------------
Date: Mon, 1 Aug 88 12:08:42 EDT
From: mike%bucasb.bu.edu@bu-it.BU.EDU (Michael Cohen)
Subject: FIRST ANNUAL MEETING OF THE INTERNATIONAL NEURAL NETWORK SOCIETY
MEETING UPDATE:
September 6--10, 1988
Park Plaza Hotel
Boston, Massachusetts
The first annual INNS meeting promises to be a historic event. Its program
includes the largest selection of investigators ever assembled to present
the full range of neural network research and applications.
The meeting will bring together over 2000 scientists, engineers, students,
government administrators, industrial commercializers, and financiers. It
is rapidly selling out. Reserve now to avoid disappointment.
Call J.R. Shuman Associates, (617) 237-7931 for information about registration
For information about hotel reservations, call the Park Plaza Hotel at
(800) 225-2008 and reference "Neural Networks." If you call
from Massachusetts, call (800) 462-2022.
There will be 600 scientific presentations, including tutorials, plenary
lectures, symposia, and contributed oral and poster presentations. Over 50
exhibits are already reserved for industrial firms, publishing houses, and
government agencies.
The full day of tutorials presented on September 6 will be given by Gail
Carpenter, John Daugman, Stephen Grossberg, Morris Hirsch, Teuvo Kohonen,
David Rumelhart, Demetri Psaltis, and Allen Selverston. The plenary lecturers
are Stephen Grossberg, Carver Mead, Terrence Sejnowski, Nobuo Suga, and Bernard
Widrow. Approximately 30 symposium lectures will be given, 125 contributed oral
presentations, and 400 poster presentations.
Fourteen professional societies are cooperating with the INNS meeting. They
are:
American Association of Artificial Intelligence
American Mathematical Society
Association for Behavior Analysis
Cognitive Science Society
IEEE Boston Section
IEEE Computer Society
IEEE Control Systems Society
IEEE Engineering in Medicine and Biology Society
IEEE Systems, Man and Cybernetics Society
Optical Society of America
Society for Industrial and Applied Mathematics
Society for Mathematical Biology
Society of Photo-Optical Instrumentation Engineers
Society for the Experimental Analysis of Behavior
DO NOT MISS THE FIRST BIRTHDAY CELEBRATION OF THIS IMPORTANT NEW
RESEARCH COALITION!
------------------------------
From: Phill Everson <everson%COMPSCI.BRISTOL.AC.UK@CUNYVM.CUNY.EDU>
Subject: Public Domain Sun Image Processing Software
Date: Tue, 19 Jul 88 13:26:39 +0100
Version 1.1.1: of ALV Public Domain (see file "COPYRIGHT") Image
Processing Toolkit for the Sun Workstation
released Sun 17 Jul, 1988
This is to introduce family of image processing programs written by
Phill Everson <everson@uk.ac.bristol.cs> with help from Gareth Waddell
(notably for the dynamic array library) at Bristol University in the UK
for SUN workstations, both colour and black/white. (The imed image
editor is largely based on sunpaint, written a year or two ago by Ian
Batten at Birmingham University. Thanks very much!) It includes tools
to display images, to convolve a filter over an image, to create a
histogram of the greylevels in an image, to perform histogram
equalisation of an image, to threshold an image, to convert an image to
Postscript and ... (read the manual page alv(1) for other features).
AlV stands for Autonomous Land Vehicle, the research project that these
were originally developed for. The toolkit was written to fullfil a
need rather than to be especially pretty, so in places there are some
rough edges. Some of the tools have been used MUCH more than
others and so can be regarded as being pretty much correct (dsp,
convolve, pixval, imagelw, subarea, subsample, winwidth, hist &
invert). If any of the others seem to be playing up it is possible
that there is a bug in there somewhere -- some tools were added at the
request of others who promised to test them and have never been heard
of since! Please send me any bug reports (and fixes please :-) ) to me.
Note that imed does *not* work with colour workstations as yet!
*************************************************************************
To get this system up and on the road:
1. Edit the Makefile, changing the directory paths for
BINDIR, LIBDIR, INCDIR, MANDIR & FILDIR to suit your system.
2. You might want to alter the command to send Postscript to
your Laserprinter in imagelw.c - at present it is "lpr -Plw -v"
on line 58.
3. Type 'make' and everything will be compiled and installed. This
takes about 15 minutes.
4. Read the manual page alv(1). It can be formatted from this
directory by typing 'nroff -man alv.1 | more'.
*************************************************************************
This family of programs has 3 manual pages; alv(1), alv(3) & alv(5).
alv(1) has a general description of each of the programs and what each
of them should do. alv(3) is a description of the library created and
alv(5) is a description of the file format used for an image. (I've
also included the manual page dynamem(3) for a dynamic memory
allocation library which is used by the alv library and which someone
may find useful.)
The method that we have found works best is that everyone working
on vision programs uses the same file format (see alv(5)) and
most people will use the core tools to display images etc and the
library functions for their own programs.
These are and will be used a lot here, so if anybody adds or modifies
them, please send me details and I'll collect, collate and send updates
out on the mailing list. It is likely that new tools will be added here
also as I'm now sure to be here until at least 1990.
If you want to be put on a mailing list for additions and bugfix
reports, mail "alv-users-request@uk.ac.bristol.cs". The actual mailing
list can be accessed by mailing to "alv-users@uk.ac.bristol.cs".
I hope they're of some use.
Phill Everson
SNAIL: Phill Everson, Comp. Sci. Dept., U of Bristol, Bristol, UK
JANET: everson@uk.ac.bristol.cs UUCP: ...mcvax!ukc!csisles!everson
ARPANET: everson%cs.bristol.ac.uk@nss.cs.ucl.ac.uk
------------------------------
Date: 28 Jul 88 12:41:17 EDT
From: Bernice Rogowitz <ROGOWTZ@ibm.com>
Subject: multidisciplinary conference on VISION, PROCESSING, and DISPLAY
********** MEETING ANNOUNCEMENT AND CALL FOR PAPERS ********
a multidisciplinary conference on:
HUMAN VISION, VISUAL PROCESSING, AND DIGITAL DISPLAY
Bernice E. Rogowitz, chairman
January 19-20, 1989
This meeting is a subset of the
SPSE/SPIE Symposium on Electronic Imaging, January 15-20, 1989
Los Angeles Hilton Hotel, Los Angeles, California
TOPICS:
o Models for Human and Machine Vision
o Color Vision and Color Coding
o Digitization, Spatial Sampling, and Anti-Aliasing
o Vision-Based Algorithms for Image Processing
o Psychophysics of Image Quality
o Spatial/Color/Temporal Interactions in Perception and Coding
CONFERENCE GOAL:
The goal of this two-day conference is to explore interactions between
human visual processing and the diverse technologies for displaying,
coding, processing, and interpreting visual information.
PARTICIPANTS:
Paper are solicited from scientists working in visual psychophysics,
computer vision, computer graphics, digital display, printing,
photography, image processing, visualization, medical imaging, etc.
IMPORTANT DATES:
150-word Abstracts Due: August 31, 1988
Camera-ready Manuscript due: December 19, 1988
FOR MORE INFORMATION, PLEASE CONTACT:
Dr. Bernice E. Rogowitz, chair SPIE Technical Program
IBM T. J. Watson Research Center 1022 19th Street
Box 218 Yorktown Heights, NY 10598 Bellingham, WA 98225
(914) 945-1687 Net: ROGOWTZ@IBM.COM (206) 676-3290
------------------------------
End of VISION-LIST
********************
