CSE/EE 576: Image Understanding
Welcome to the 576 Home Page!
This is the World Wide Web ("the Web" for short) hypermedia document
for CSE/EE 576 and contains information about the
class. Keep in mind that this document is not static, and that new
information (especially class messages) will be added frequently.
If you have any problems with this document, send mail to
Copyright Notice: The material in this course web is subject
to copyright. While it may be viewed by the public,
it should not be installed at any web site other
than the one at the University of Washington.
The first assignment is to read Chapter 1 of the course notes and do
The second assignment is due Wednesday, April 5.
Read Chapter 2 of the course notes and do exercises 1-7.
Next, determine the most convenient way to
run KHOROS and CANTATA for you. In CSE it runs on Suns such as
Edit your local workstation's .login file to contain
Then add to you .cshrc file on the Suns the following:
After an rlogin onto Lillith with your rhost assignments
and DISPLAY environment variable set up appropriately, type
cantata at the Unix prompt. In EE, the machine "george" has
KHOROS and CANTATA installed on it.
Also go to our course home page on the WWW
and follow the links to the KHOROS/CANTATA tutorial. Go to its
course outline, to "experiments" and do at least the first two
topics ("Image information" and "Spatial Resolution").
The ideal way to take the tutorial is to read the web pages
and experiment with KHOROS in another X window.
There is nothing to turn in for this part of the assignment.
- setenv KHOROS_HOME /usr/local/khoros
- setenv MANPATH /usr/local/khoros/man
- set path=($KHOROS_HOME/bin $path)
The third assignment is to read the article by Lin, Huertas, and
Nevatia on pp.62-69 of the CVPR'94 proceedings, and also to
look at the article by Wolff on pp.369-376.
The fourth assignment is due Monday, April 17. In this assignment,
we make a comparison of three image-processing software environments:
KHOROS, MSVC/C++ & "Image", and MATLAB Image Processing Toolbox.
The objects of the assignment
are (1) to gain some experience with each environment,
(2) to begin to get a feel for the strengths and weaknesses of
each of these environments, and (3) to explore the use of
convolution in noise suppression and edge detection.
In each environment: (a) select a monochrome test image.
(b) apply the Sobel operation to the test
image. (c) apply a moving average filter to the test image.
(d) apply a 3 by 3 laplacian to the test image.
(e) simulate gaussian smoothing with larger standard-deviation
gaussians by iteratively applying the moving average filter.
(f) apply the 3 by 3 laplacian to each gaussian-smoothed image.
(g) describe the results obtained in terms of changes in appearance
to objects and contours in the test image.
(h) describe the following aspects of the implementation:
what menu selections, direct manipulations, or programming you
had to do to perform the experiments; how fast the operations ran;
and level of learning effort required on your part.
The final examination will be on Monday, June 5 from 2:30 to 4:20PM
in our regular class meeting room. The exam will cover a combination
of pre-midterm and post-midterm material.
Here is a list of topics to study.
Reminder: the class approved Sunday, June 4 from 1:30-2:30
as the time for the final-exam review meeting. The plan is to meet in
Outlines for selected class periods are available here:
Friday, 28 April.
Monday, 1 May.
Wednesday, 3 May.
Monday, 8 May.
Friday, 12 May.
Monday, 15 May.
Wednesday, 17 May.
Friday, 19 May.
Monday, 22 May.
Wednesday, 24 May.
Copies of the overhead transparencies for the May 22 and 24 lectures on
neural nets are available at the Engineering Library Copy Center (2nd floor)
as CSE/EE 576 packet number 2 ("Trainable Classifiers").
Friday, 26 May.
Students in 576 are permitted to get a temporary copy of
MATLAB for their own use in the course. This will require
that you fill out a form and sign a contract.
Let me know if you are interested.
Term projects are an important part of
the course. These are to be started during the week of April 24.
Here are some corrections to the course notes.
On Friday, April 7, we were introduced to the Pentium laboratory, including
the MSVC/C++ software development environment.
Here is Evan McLain's document
explaining how to add a new transform to the Image application.
Current status of the KHOROS installation on the Suns.
It contains the most recent information about getting started with
Khoros and the accompanying DIP course.
Please read this.
No class on Friday, March 31 or Monday, April 3.
Most students already have a computer account that can access
KHOROS and CANTATA. If you do not, contact Renee Reed on
Thursday or Friday to make an arrangement to pick up your
account login name and password so that you can use it
over the weekend and early next week. Renee's hours are
limited, so plan ahead. Her email address is reed@cs, and
she is in part of Sieg 127 (in a back office).
Sieg 127 is kept locked, so you will either need to knock or
have a prior arrangement to meet her.
Many of the account application cards were not signed. We will
have to take care of that next week.
Selected Lecture Slides
Monday, March 27.
Wednesday, March 29.
Information Resources for Image Understanding
An Online Intro to Image Processing with KHOROS/CANTATA
The Delft Univ. Pattern Recognition Information Page.
The Computer Vision Home Page (stored at CMU).
If you're running Mosaic on a Mac, you can find some help items
under the Mac Balloon Help menu, and on the MacMosaic Home Page item
under the Navigate menu, and
(Last Update: 31 May 1995)