bobk@boulder (Bob Kinne) (06/09/89)
The Engineering Research Center for Optoelectronic Computing
Systems at University of Colorado, Boulder, will present two
short courses this summer.
Optoelectronic Computing Architectures - July 26-28
Optical Processing for Pattern Recognition and Image
Understanding - July 31-August 2.
_______________________________________________________________
Course Instructors
W, Thomas Cathey. PhD. Dr. Cathey, Professor at UCB, has worked
in the fields of optical signal processing, optical computing, and
optical systems since 1964. He is presently the Director of the
Optoelectronic Computing Systems Center.
Karl-Heinz Brenner, PhD. Dr. Brenner, Head of the Optical Computing
Group at the University of Erlangen in West Germany, has conducted
research in the field of optical symbolic substitution at Bell
Laboratories and the University of Erlangen and is writing on the
subjects of symbolic substitution and optical computing.
Mark O. Freeman, PhD. Dr. Freeman, Assistant Professor with
the OCS Center, is performing research in optical signal processing
and optical systems for image understanding. His current projects
include optical scale-space methods for image analysis and optical
geometrical transformations.
Vincent Heuring, PhD. Dr. Heuring, Associate Professor at CU-Boulder,
is responsible for the interface and software for the optical bit
serial computer under construction at the Optoelectronic Computing
Systems Center.
K. S. Huang, PhD. Dr. Huang, Research Staff Member in the Computer
Science Department of IBM in Yorktown, is one of the developers of
binary image algebra.
William T. Rhodes, PhD. Dr Rhodes, Professor of Electrical Engineering
at Georgia Tech, is also the editor of Applied Optics. He is
responsible for many useful techniques in optical signal processing
and is currently pursuing the use of these systems for morphological
image processing.
Kelvin Wagner, PhD. Dr Wagner, Assistant Professor of Electrical
Engineering at CU-Boulder, is performing research on optical neural
networks, optical data storage, and is completing a book on the use
of acousto-optic devices for signal processing.
______
OPTOELECTRONIC COMPUTING ARCHITECTURES
This course will provide a discussion of the potential advantages
of optoelectronic computing, the near-term applications, and the
impact on future computing systems. The course will cover optical
computing architectures that either have been or are being built.
Appropriate attendees include engineers, scientists, and technical
managers with backgrounds or interests in computer architectrues,
optical computing, and their applications.
Course Content
WEDNESDAY AM - Introduction and Background - Cathey
A review of previous research in optical computing and the impact
of optics on computing architectures.
Wednesday PM - Optical Bit Serial Computing - Heuring
A description is given of the architecture and design of a stored
program, bit serial optoelectronic computer that is constructed with
lithium niobate directional couplers and optical fibers.
THURSDAY AM - Symbolic Substitution - Brenner
The theory of the use of symbolic substitution in two-dimensional
optical computing systems is presented. Proposed architectures and
implementation techniques are described.
THURSDAY PM - Binary Image Algebra - Huang
A presentation of the basic concepts of binary image algebra systems
including: how they serve as a unified theory leading to a formal
parallel language for two-dimensional processing, how optical
architectures can implement binary image algebra algorithms, potentials
and limits, and experimental results.
FRIDAY AM - Optical Programmed Logic Arrays - Rhodes
The architecture is given of a parallel computing system that is
based on the use of optical programmed logic arrays. Various
implementations of subsystems are presented.
FRIDAY PM - Optical Neural Nets - Wagner
Simple nonlinearities and global interconnectivity provided by
optical devices lead to natural hardware implementations of neural
networks. Architecture based on spatial light modulators and
dynamic holographic materials will be presented, as well as the
capabilities and limitations of this approach to computation.
Advantages, limitations, and examples of optical implementations of
neural-like architectures will be discussed.
_____
OPTICAL PROCESSING FOR PATTERN RECOGNITION AND IMAGE UNDERSTANDING
The course begins with a review of fundamental optical operations and
covers the development of optical vision processors up through the
current state-of-the-art. It is designed to introduce both
mathematical reasoning behind the various image operations and
optoelectronic systems that can perform these computations. Advantages
and limitations of optical implementations will be discussed. This
course is intended for engineers and managers with an interest in
automated vision systems for inspection, pattern recognition, tracking,
and other such tasks.
Course Content
MONDAY
Introduction to Optical Processors - Freeman and Rhodes
The course begins with a brief look at some of the fundamental
properties of optical systems which make them attractive for processing
images. A major emphasis is on space-invariant optical processors
using both coherent (Freeman) and incoherent (Rhodes) light.
Image Conditioning and Low-Level Understanding - Rhodes
Often an input image contains noise or distracting information. This
section is directed toward methods for noise removal and segmentation
using threshold decomposition and morphological image operations.
TUESDAY
Image Conditioning and Low-Level Understanding - Freeman
(Continued from Monday). Optical techniques for attention focusing
and object tracking are presented.
Computer-Generated Holograms - Freeman
Computer-generated holograms allow one to create somewhat arbitrary
complex field distributions. A survey of coding techniques and
practical considerations are introduced.
Optical Pattern Recognition I: Relatives of Matched Filters - Freeman
This section summarizes the numerous optical pattern recognition
approaches which are tied to matched filters that have been published
in recent years. Topics include: phase-only matched filters,
invariant-matched filters, synthetic discrimination functions, and
lock and tumbler filters.
Optical Pattern Recognition II: Feature-Based Recognition - Freeman
Here another approach to pattern recognition is considered; object
classes are characterized by a set of features. General features not
tied to any particular application as well as optimum features for
specific applications are discussed. A method to compute the
necessary inner products optically is also presented.
WEDNESDAY AM
Optical Pattern Recognition III: Neural Network Approaches - Wagner
The final pattern recognition approach to be considered is based on
neural network techniques. Fundamental concepts in neural networks
are introduced followed by a number of specific topics including:
single and multiple layer networks, pattern recognition systems, and
optical implementations.
_______________________________________________________________
These courses will be held in the Engineering Center at the
University of Colorado, Boulder, Colorado.
Course fees include materials, refreshments, and a barbeque.
Upon registration you will receive a packet containing information
regarding lodging and about the city of Boulder.
For further information, call (303) 492-7129 or (303) 492-2299.
Registration is due by Friday, July 14, 1989.
Enrollment form:
---------------------------------------------------------------
I wish to enroll in:
___ Optoelectronic Computing Architectures - $750
___ Optical Processing for Pattern Recognition and Image
Understanding - $750
___ Both of the above courses - $1300
Name: ___________________________________________
Job Title: ______________________________________
Company: ________________________________________
Address: ________________________________________
________________________________________
Telephone: ______________________________________
Please return registration form and tuition fee to:
Kathleen Exman, Conference Coordinator
University of Colorado at Boulder
Optoelectronic Computing Systems Center
Campus Box 525
Boulder, Colorado 80309-0525
Payment or purchase order must be sent prior to the start of
the course. Checks should be made payable to: Optoelectronic
Computing Systems Center.