Copy Link
Add to Bookmark
Report
VISION-LIST Digest Volume 11 Issue 10
VISION-LIST Digest Mon Mar 23 12:50:09 PDT 92 Volume 11 : Issue 10
- 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:
Sorry for strangenesses on Vision List Archive
Waltz' algorithm
Simulated camera in robotic cell
Automatic Texture Generator
Three object triangulate problem
Geometric modeling software request
Representing/encoding hierarchies
Neural networks for classification
CIE to RGB routines
Testing of model-based object recognition systems
USDA National needs Fellowship at the University of Illinois
Job sought...
Aug 92 Washington DC: Photogrammetry & Comp. Vis. (long)
----------------------------------------------------------------------
Date: Thu, 12 Mar 92 16:33:24 -0800
From: vision (Vision-List-Request)
Subject: Sorry for strangenesses on Vision List Archive
The Vision List site has been reworking the topology and file system (to AFS).
In addition, the anonymous ftp site was changed to FTP.ADS.COM in order
to better segment our system for security reasons. Please let me know if you
experience any problems.
Some things to remember:
1) Ftp to site FTP.ADS.COM . If your site does not correctly support this
address, you will need to ask me at vision-list-request@ads.com for the
absolute IP address for the archive site: the same IP address it is not
guaranteed to remain the same forever, though the logical name
FTP.ADS.COM will remain stable.
2) The command "dir" is the accepted way to list directories in ftp. There
have been some problems reported with the use of "ls" within ftp that
are AFS related. No big problem, but you need to know about it (it is
being worked on). "dir <directory>" is currently being fixed.
3) LET ME KNOW OF ANY PROBLEMS YOU HAVE SO THAT I CAN INVESTIGATE AND REPAIR.
Sorry for any inconvenience,
phil...
------------------------------
Date: 5 Mar 92 10:18:39 GMT
From: baustad@idt.unit.no (Jostein Baustad)
Organization: Norwegian Institue of Technology
Subject: Waltz' algorithm
I am working on a project where I am trying to analyze a line drawing
of a scene of objects, for instance cubes. I am basing my project on
the work done by D.L. Waltz described in "Understanding line drawings
of scenes with shadows" in "The Psychology of Computer Vision", ed. P.
Winston. This articlecollection is from 1975 (!) and I am pretty sure
there's been done a lot of work on this topic since then.
Can anyone point me to related work ? (ftp-sites are ok). Is Waltz' PhD thesis
available on any ftp-sites ?
Please e-mail me directly since I am not reading these groups regularly.
Thank you in advance.
* Jostein Baustad * E-mail : baustad@idt.unit.no *
* Norwegian Institute of Technology * baustad@solan.unit.no *
* Division of CS and Telematics * "Reality is just a convenient *
* Information and Knowledge Systems * measure for complexity." *
------------------------------
Date: Wed, 11 Mar 92 09:23:23 +0100
From: benamar@lai1.univ-lyon1.fr (Chokri BENAMAR)
Subject: Simulated camera in robotic cell
Would somebody point out to me some recent papers or some examples in
CCD-simulated cameras and their integration in a simulated robotic
cell. I'd like to use vision geometric data downloaded from a
simulated camera system to program robot motion.
sincerely,
* BEN AMAR Chokri *
* Laboratoire d'Automatique Industrielle *
* bat 303, 20 avenue Albert Einstein *
* 69100 VILLEURBANNE *
* tel. : (33) 72.43.81.98 *
* fax. : (33) 72.43.85.15 *
* e-mail : benamar@lai1.univ-lyon1.fr *
------------------------------
Date: Mon, 9 Mar 92 21:48:47 KST
From: dmyoon@lion.postech.ac.kr (Yoon Doo Man)
Subject: Automatic Texture Generator
Hi,
I'm going to test my pattern classification algorithm which
would be a part of automatic visual inspection system.
But major problem to test it is that I can't get enough test images.
I'm trying to test my texture classification algorithm with massive
texture images. I need a hundred textures for each class. And I want
to get 100 classes ---- 10000 images in total. I think I can't get such
a massive images files from anyone . So what I want to is an automatic
texuture generation program which can generate various texture images
with different parameters.
So I'm trying to find the automatic texture generator
which generates variant classes on different parameters.
Would you send me your source code or execution code for it
if you have ?
Thanks in advance !
E-mail :dmyoon@csd.postech.ac.kr
Doo-man Yoon Tel : 0562 - 79 - 2916
FAX : 0562 - 79 - 2299
Dept. of computer science
Pohang Institute of Science and Technology
P.O.BOX 125
Pohang,Kyeongbuk KOREA
------------------------------
Date: Sun, 15 Mar 92 21:43:57 EST
From: charles@caen.engin.umich.edu (Charles Jacob Cohen)
Organization: University of Michigan Engineering, Ann Arbor
Subject: Three object triangulate problem
I need help with a problem. I have a solution, but it fails in certain
cases. Anyway, given:
Object one at x1, y1
Object two at x2, y2
Object three at x3, y3
A robot at an unknown x,y, and it has an orientation error theta.
Also, given o1, o2, and o3.
o1 is the angle formed by measuring an angle counter-clockwise from the
positive x axis to the line connecting object one to the robot. This
is measured from the robot, and therefore *includes* the orientation error.
o2 and o3 are measured the same way, but to objects 2 and object3.
Problem: What are the true x and y coordinates of the robot, and its
orientation error?
I'm sure this *has* to have been done before, but I can't find any
papers on it. If desired, I will post my solution, which usually works,
but not all the time. Thanks for your time - Chuck Cohen
------------------------------
Date: Mon, 16 Mar 1992 13:38:42 GMT
From: petern@bacall.bion.kth.se (Peter Nordlund)
Organization: Royal Institute of Technology, Stockholm, Sweden
Subject: Geometric modeling software request
I am looking for public domain software for geometic modeling, to be
used in a model-based recognition system. This sysem is supposed to
deal with a few known man-made objects. C language,UNIX and X window
versions are preferred. Any info will be appreciated.
Thanks.
My address is:
Peter Nordlund
KTH (The Royal Institute of Technology), NADA, CVAP
100 44 Stockholm
SWEDEN
tel: +46-8-790 69 06
fax: +46-8-723 03 02
e-mail: petern@bion.kth.se
------------------------------
------------------------------
Date: Fri, 20 Mar 92 13:35 GMT
From: GLYNN PETER ROBINSON <GROBINSON@portia.umds.lon.ac.uk>
Subject: Representing/encoding hierarchies
Can anyone provide me with any references on techniques for
representing/encoding hierarchies (particularly ones defined on
images).
Thanks in anticipation,
Lewis Griffin
------------------------------
Date: Wed, 4 March 92 16:50:00 GMT
From: iborra@disam.upm.es
Subject: Neural networks for classification
We are involved in a real time classification problem in which it is
necesary t o classify about fifty different defects detected onto a
flat surface.
Have anyone used a neural network for classification proposals ?
What we want to know is if is there any commercial product
(development softwar e + network board) adecuated for such
application. We would like to have more detailed information like
company name, price, real-time capabilities,...
Your help would be highly appreciated.
------------------------------
Date: Thu, 12 Mar 92 14:18:26 CST
From: braker@shadow.brooks.af.mil (Jim Brakefield MSEE)
Subject: CIE to RGB routines
I'm trying to implement CIE to RGB conversion. I have typed in the
implied math from pg. 348-350 of Fundamentals of Three-Dimensional
Graphics by Alan Watt, Addison-Wesley 1989. It doesn't quite work
(its off by some scaling factor). Is there any Fortran, C or Pascal
code out there that does this. Input is CIE x y & Y of the red, green
and blue phosphours. Output desired is a matrix that takes a given
X Y & Z and converts it to RGB brightness levels.
Thanks
Jim Brakefield 512-533-1228
KRUG Life Sciences, San Antonio, Texas
------------------------------
Date: Sun, 22 Mar 1992 01:26:23 -0500
From: "Chennubhotla C. S." <chakra@iris.psu.edu>
Subject: Testing of model-based object recognition systems
Subj: Suggestions needed on creating random 2D intensity/3D range images
for testing model-based object recognition systems.
As I see, research in object recognition (2D similarity, 2D affine, 3D
from 2d and 3D range) has been going on for atleast 2 decades now.
What I dont see is some kind of "benchmark" to let me characterize
various algorithms and their design choices. My attempt at creating
random images is on these lines. I would like to have your suggestion
on the following:
1) did someone look into this idea of random images already?
Some of the references I found [1], [2] fall short of being benchmarks.
2) Is it worth the effort? is there a constraint in using only those models
which are in the database and not being able to "create" random ones?
Well, if it is worth it, then...
3) what are the typical values for the various input parameters?
To make the random scene look real, what do the more experienced
researchers in object recognition have to say on the following:
a) what is a reasonable limit to the number of models in the scene?
b) how much is typically the size of a model?
c) how much is typically the occlusion?
d) how much noise should be introduced in the scene?
e) given that "lighting" and "surface characteristics" can play
a major role, is it possible to set some kind of limits on this?
f) what is the range on "viewing" parameters?
g) what parameters are crucial in deciding the performance of your
recogniton system?
h) are there any other parameters of interest?
4) where can I find models for the database?
For 3D range imagery I know of one such database in public domain.
Is there one for 2D images too?
REFERENCES:
[1] Jacobs, D. W.-The use of grouping in visual object recognition -
MIT AI Lab TR 1023.
[2] Saund, Eric-Role of Knowledge in the visual shape representation -
MIT AI Lab TR 1092.
If there is sufficient interest, I will definitely summarize the postings.
thank you for your time...
chakra
(chakra@iris.psu.edu)
------------------------------
Date: Fri, 20 Mar 92 13:21:44 CST
From: "John F. Reid" <JFREID@vmd.cso.uiuc.edu>
Subject: USDA National needs Fellowship at the University of Illinois
AREA OF RESEARCH: MACHINE VISION SENSING FOR MEASUREMENT AND
CONTROL OF BIOLOGICAL SYSTEMS
DESCRIPTION:
Machine vision systems offer great potential for sensing the physical
properties of biological products. Biologically related properties
can be difficult to quantify. When applicable, computer vision sensing
results in improvements in efficiency, energy utilization and product
quality.
A key component of a machine vision system is determining the important
parameters to sense and then detecting those parameters accurately and
efficiently. The focus of the Fellow's work will be to develop an
understanding of areas critical to the development of a vision sensor
including the spectral response; image processing and feature extraction
techniques that can be used to isolate the biological response;
pattern recognition techniques that can be used to make a decision from
the detected response; and control actions that result from a sensed
phenomena.
CURRENT RESEARCH:
Qualified research fellows will become part of a successful
research program in machine vision applications:
-Vision-based Sensing for the Control of Fermentation Processes
-Sensing Control Parameters for in vitro Production
of Natural Plant Pigments
-Real-time Machine Vision For Grain Quality Inspection
-Machine Vision Measurement of a Steeping Index for
Wet-milling of Corn
-Calibration of Color Machine Vision Systems
REQUIREMENTS:
U.S. Citizens with B.S. and M.S. At least one degree must be
in Engineering.
STIPEND:
$17,000 per year for 3 years plus tuition and fees waiver.
INFORMATION:
For more information, please contact:
Dr. John F. Reid
Agricultural Engineering Department
University of Illinois
Urbana, IL 61801
Phone: (217) 333-2738
FAX: (217) 244-0323
e-mail: reidj@cardamom.age.uiuc.edu
------------------------------
Date: Tue, 10 Mar 1992 18:13:00 GMT
From: cmcl2!acf5.NYU.EDU!mcb9594@uunet.UU.NET (M.Biondi)
Organization: New York University
Subject: Job sought...
NYU Computer Science May graduate looking for full time job in the Computer
Science or Robotics Fields. Extensive experience with 8088 and 6502
family assembly languages, C, C++, and Pascal. Have taken courses in
Operating systems design, Computational Theory, Robotic Theory, and CAD/CAM.
Also familiar with manipulator dynamics, forward and inverse kinematics,
force control, and trajectory generation.
I hope I am the correct solution to your problems.
Michael Biondi
------------------------------
Date: Wed, 11 Mar 92 19:50 MET
From: "Herman J. Woltring" <UGDIST@nici.kun.nl>
Subject: Aug 92 Washington DC: Photogrammetry & Comp. Vis.
Dear Vision-List readers,
The item below was originally distributed today via Biomch-L@nic.surfnet.nl,
an email list on Biomechanics and Movement Science.
N.B.: The term "calibration" in photogrammetry usually refers to the so-called
interior camera parameters -- principal point, principal distance, image skew-
ness, differential scaling between image axes, and non-linear image distortion
parameters. The term "orientation" usually refers to the *pose* of a camera,
i.e., both position and attitude with respect to a chosen coordinate system.
Conversely, the biomechanics and kinesiology field usually takes "calibration"
to encompass both "interior camera" and "exterior orientation" parameters,
while the term "orientation" usually refers to attitudes only.
Herman J. Woltring, Eindhoven/NL
P.S.: An Email contact for this meeting is currently being investigated;
interested readers may drop me a note.
WORKSHOP UT-1,
"Calibration and Orientation of Cameras in Computer Vision"
Date: Sunday, 2 August 1992, 09:00 - 19:30
Venue: 17th ISPRS Congress
International Society for Photogrammetry and Remote Sensing
Washington DC, 2-14 August 1992
Fee: US $ 200,00 including lecture notes and refreshments
(US $ 100 for students)
Co-chairs: A. Gruen, Institute of Geodesy and Photogrammetry,
Federal Institute of Technoloy / ETH Zuerich, Switzerland
T.S. Huang, Coordinated Science Laboratory and Beckman Institute
University of Illinois at Urbana-Champaign, USA
Objective: In recent years many researchers in computer vision have realized
the importance of accurate calibration and attitude determination in stereo-
pose determination, 3D motion estimation, and related tasks. A number of
these researchers have investigated the problem and obtained significant
results. However, the problem of camera calibration and orientation determi-
nation has been studied by photogrammetrists for many years (starting long
before the field of computer vision was born). Many of the computer vision
researchers are thus unkowingly reinventing the wheel. On the other hand, some
of the results obtained in computer vision, including the minimal information
required for unique solutions, degenerate configurations, and efficient linear
algorithms may be new and useful for photogrammetrists.
Therefore, the objective of this Workshop is to bring together a small group
of researchers in photogrammetry and computer vision so that they can exchange
information and foster future collaborative efforts.
Program:
09:00 - 10:30 R. Wrobel, Institute of Photogrammetry and Cartography,
Technical University Darmstadt, Germany: Minimum solutions
for orientation.
T.S. Huang, University of Illinois at Urbana-Champaign, and
R. Tsai, IBMT.J. Watson Research Center, USA: Uniqueness,
number of solutions and degeneracy in determining camera
orientations.
11:00 - 12:30 W. Foerstner, Institute of Photogrammetry, University Bonn,
Germany: Robust orientation procedures with minimum and
redundant information.
D. Gennery, Jet Propulsion Laboratory, Pasadena, CA, USA:
Least-squares camera calibration including lens distortion
and automatic editing of calibration points.
14:00 - 15:30 C. Fraser, Geodetic Services Inc., Melbourne, Florida, USA:
Camera component calibration techniques - theory, systems,
results.
S. Shafer, Carnegie-Mellon University, Pittsburgh, PA, USA:
Camera calibration for high-precision machine vision.
16:00 - 17:30 A. Gruen, H. Beyer, Institute of Geodesy and Photogrammetry,
ETH Zuerich, Switzerland: System calibration through self-
calibration.
O.D. Faugeras, INRIA - Sophia Antipolis, France: Automatic
calibration of a stereo rig
18:00 - 19:30 Summary discussion
Registration: Those attending the ISPRS Congress should use the related
Advance Registration Form. Those attending only the Work-
shop should use the special TU-1 Registration form, to be
obtained from:
XVII ISPRS Congress, c/o Galaxy Registration
P.O. Box 4088, FREDERICK, MD 21701, USA
Fax: +1.301.662-9411 (mention: event # 69)
This Workshop is cosponsored by ISPRS Commission V and IEEE Computer Society
Technical Committee on Pattern Analysis and Machine Intelligence (PAMI).
------------------------------
End of VISION-LIST digest 11.10
************************
