[191 students: This talk will be aimed at faculty and graduate students,
so it will likely assume some background that you don't have yet.  But
you may find it interesting because it involves both quad-trees and
Huffman coding. --Diane]


-------------------------------------------------------------------------------
University of Toronto
Edward S. Rogers Sr. Department of Electrical and Computer Engineering
Distinguished Lecture Series
Presents

"Rate-Distortion Techniques in Video Processing and Communications"

Aggelos K. Katsaggelos
Professor
Ameritech Chair of Information Technology
Northwestern University
Department of ECE
Evanston, IL 60208-3118

Date:  Tuesday, March 6, 2001
Time:    3:00 p.m.
Place:    Sandford Fleming Building, Room SF1105

ALL FIRST YEAR ECE MASTER'S AND PH.D. STUDENTS MUST ATTEND

Abstract

In this talk the application of rate-distortion techniques
to video processing and communications is addressed.
As an example, the rate-distortion optimal bit allocation among
quad-tree-based segmentation, displacement vector field and
displaced frame difference is presented and demonstrated.
As another example, the problem of the efficient encoding of
object boundaries in the rate-distortion sense is described in detail.
This problem is becoming increasingly important
in applications such as content-based storage and retrieval, studio
and television post-production and mobile multimedia applications.
MPEG-4 Visual, for example, will allow the transmission of arbitrarily
shaped
video objects, while MPEG-7 is using shape as one of the visual
descriptors.
The object boundaries are given as binary masks,
and they are approximated by a curve of a given order, such as, a
polygon
or a B-spline. Various criteria are considered for measuring the
distortion between a segment of the original curve and its
approximation,
as well as, the total distortion.  For all these distortion criteria,
and a predetermined way for defining the rate,
the problem of finding the approximation to a given boundary
which results in the smallest distortion for a given number of bits is
solved. The approach is extended to include the simultaneous
optimization
of the (Huffman) code table, as well as, inter-object and
inter-frame boundary encoding.
Finally, the rate-distortion formulation and solution of other
multimedia communication problems will be considered.

----------

Short Biography

AGGELOS K. KATSAGGELOS is the Ameritech Professor of Information
Technology in the Department of Electrical and Computer Engineering
at Northwestern University, and Director of the
Northwestern-Motorola Center for Telecommunications.
He is also a member of
the Associate Staff, Department of Medicine, at Evanston Hospital.
He received the Diploma degree in electrical and mechanical engineering
from the Aristotle University of Thessaloniki, Greece, in 1979,
and the M.S. and Ph.D. degrees both in electrical engineering
from the Georgia Institute of Technology in 1981 and 1985, respectively.

He has published extensively in the areas
of image and video recovery, video processing and compression,
and multimedia signal processing.
He is the editor of "Digital Image Restoration", (Springer, 1991),
co-author of "Rate-Distortion Based Video Compression", (Kluwer, 1997),
and co-editor of "Signal Recovery Techniques for Image and
Video Compression and Transmission", (Kluwer, 1998).
He is a member of the Board of Governors and the
Publications Board of the IEEE Signal Processing Society,
the IEEE TAB Magazine Committee, and Editor-in-Chief of the
IEEE Signal Processing Magazine.
He is Fellow of the IEEE and recipient
of the IEEE Millennium Medal.