DEPARTMENT OF COMPUTER SCIENCE

                            COURSE ANNOUNCEMENT

                      Introductory Graduate Course on
                             VISUAL PERCEPTION

                              COMP 290-050
               Variable Credit (see instructor for details)
              Tuesday and Thursday, 12:30-1:45 Sitterson 325
                    Instructor: Dr. Jonathan Marshall

Visual perception is one of the most active, challenging, and exciting
areas of scientific research today.  The field is concerned with
understanding how the light that bounces around our world gets transformed
into the vivid, informative experience that we call vision.  The field is
highly interdisciplinary, involving psychophysics, neurophysiology, and
computational models.

This graduate course will survey the behavioral, neurobiological, and
computational aspects of visual perception in humans and animals.  We will
cover the following topics:

  o depth perception: how we see in 3-D;
  o motion perception and visual event perception;
  o shape and form perception;
  o recognition and definition of objects;
  o receptive fields, stimulus preferences, and visual context;
  o boundary, orientation, and curvature perception;
  o perception of color and brightness;
  o texture perception;
  o visual segmentation;
  o perception of transparency, shadows, and specularities;
  o psychophysics of vision;
  o physiology and architecture of the retina-geniculate-cortex pathway;
  o representation, coding, and models in vision;
  o multiresolution and multiscale image representations;
  o parallel pathways for processing of visual information.

The course will be of interest to students with many different research
backgrounds and plans, particularly in computer science, neurobiology, or
psychology.  Besides students of perception per se, the course should also
be of interest to students in computer graphics, image analysis, or
computer vision, because it will help those students understand (1) how to
make humans and computers interact more effectively, (2) how to build
better computer vision systems by using principles in common between human
and computer vision, (3) how to build better visualization tools, and
(4) how to build better displays according to principles of "inverse
graphics."

The seminar workload will consist of readings (primarily from books,
supplemented with some papers from research journals), class discussions
and participation, a project or paper and possibly an exam.  Books:

  o (Required) Wandell, Foundations of Vision: Behavior, Neuroscience,
               and Computation, manuscript (course packet), 1993.
  o (Recommended) Sekuler & Blake, Perception, 3rd Edition, 1993.

The first 1/3 of the course will coincide with the first 1/3 (the "vision"
part) of Comp 235, "Images, Graphics, and Vision" and will be taught by
Prof. Stephen Pizer.  The latter 2/3 of the course will be taught by
Prof. Marshall.  Students may register for either this course or Comp 235
for 3 credits, or for both courses for a total of 5 credits.  With
permission of the instructor, students may register for the latter 2/3 of
the course only.  The registration call number for Comp 290-050 is 91456.
The registration call number for Comp 235 is 91449.

Prior exposure to material on vision is not required.  Graduate students in
other departments, as well as those in computer science, are encouraged to
take the seminar.  Students in Cognitive Psychology may count the course
toward their requirement of 4 courses in cognitive psychology.  Faculty
members and postdocs are also welcome.

For more information, contact Dr. Jonathan Marshall at (919) 962-1887
(marshall@cs.unc.edu).