526 Computer Networks and Distributed Systems.
3 Credits.
Introduction to computer networks and distributed systems; data
communications protocols; local area networks; distributed
applications;
distributed systems architecture and design
goals; interprocess communication and synchronization; concurrent
programming with threads; client-server programming (with Berkeley
sockets); distributed applications development using remote procedure
calls.
Text: UNIX Network Programming, W. Richard Stevens, Prentice Hall,
Englewood Cliffs, New Jersey, 1990.
Instructor: Satyendra P. Rana.
586 Introduction to Pattern Recognition and Image Processing.
3 Credits.
Model of a pattern recognition system;
representation techniques for classifiers; parametric and
nonparametric classification methods; clustering; fundamentals of
image formation and acquisition; image enhancement methods; feature
extraction for two-dimensional visual pattern recognition; document
image processing and recognition.
Text: Digital Image Processing, 2nd Edition, Rafael C. Gonzalez and Paul Wintz,
Addison-Wesley, Reading Massachusetts, 1987.
Instructor: Jia-Guu Leu.
662 Matrix Computation I.
4 Credits.
Background matrix algebra; linear system
sensitivity; basic transformations; Gaussian elimination; symmetric
systems; positive definite systems; Householder method for least
squares problems; unsymmetric eigenvalue problems; the QR algorithm.
Text: Fundamentals of Matrix Computations, David S. Watkins,
John Wiley and Sons, New York, 1991.
Instructor: Anthony Chronopolous.
726 Distributed Systems II.
3 Credits.
Design issues of
distributed systems; distributed synchronization and resource
allocation algorithms; distributed file systems; transactions in a
distributed system; distributed object management.
Text: UNIX Network Programming, W. Richard Stevens, Prentice Hall,
Englewood Cliffs, New Jersey, 1990.
Instructor: Satyendra P. Rana.
762 Matrix Computation II.
3 Credits.
Special linear
systems; Givens and fast Givens methods for least squares problems;
symmetric eigenvalue problems; singular value decomposition; Lanczos
methods; iterative methods for linear systems; functions of matrices.
Text: Matrix Computations, 2nd Edition, Gene H. Golub and Charles F. Van Loan,
Johns Hopkins University Press, Baltimore, 1989.
Instructor: N.K. Tsao.
785 Artificial Neural Networks.
3 Credits.
Introduction to computation characteristics of the brain;
single layer neural nets; multilayer nets; learning and
self-organization; adaptive and associative neural processing; current
implementations and applications.
Text: Introdution to Neural Networks, Jacek M. Zurada, West Publishing Company,
St. Paul, 1992.
Instructor: Ishwar K. Sethi.
786 Computer Vision.
3 Credits.
Low-level vision
processing; use of constraints in vision processing; three-
dimensional object recognition; dynamic scene analysis; model-based
vision systems; use of neural and fuzzy logic methods in vision.
Text: Computer Vision, Edited by Rangachar Kasturi and Ramesh C. Jain,
IEEE Computer Society Press, Los Alamitos, California, 1991.
Instructor: Jia-Guu Leu.
790 Directed Study -
Direct Analysis of Three Dimensional Bone Structure from
Three Dimensional Microtomography Image Data.
3 Credits.
The main objective of this project is to measure parameters
characterizing microscopic bone structure directly from three
dimensional (3D) image data acquired from an x-ray microtomography
system designed to image bone biopsy specimens. This will be
accomplished by extending common 2 dimensional (2D) image texture
analysis methods to 3 dimensions. The current methods for assessing
bone structure involve so called stereologic methods applied to thin
planar sections of bone specimens. The stereologic methods are time
consuming and destroy the specimen, particularly the connectivity in
the direction perpendicular to the planar slice. The loss of
information in the third dimension is avoided in direct 3D measures and
is significant for measuring the anisotropic structures found in bone.
The microscopic structure of bone is important in assessing bone strength and diagnosing disease. Bone material is constantly changing. Over the course of 4-8 months, the entire bone surface is eroded and refilled. The total amount of bone material decreases with age, which can alter the microscopic structure. Other diseases, such as osteoporosis and renal failure can also decrease the total amount of bone material.
Detailed investigations are routinely made for 2D sections of bone. Because of the destructive nature of the 2D sectioning, information regarding bone distribution perpendicular to the section cannot be fully measured. The texture features most commonly measured include bone volume, total volume, surface area, and Euler number. The 3 dimensionality of these measures is inferred by assuming an isotropic structure (which is not always true). Other texture features might be useful in structural classification, such as moments, Fourier spectrum, autocorrelation, and co-occurrence matrices.
While the main application for this project is microscopic bone specimens,
the texture features measured should be directly applicable to other
3D image data, such as whole body tomography and 3D microscopic analysis
of other materials.
Advisor: Jia-Guu Leu.
790 Directed Study - Ring Reduction in Computed Tomography.
3 Credits.
A method was investigated for reducing the circular artifacts in cone beam
tomographic images due to sensor nonuniformity is presented.
The reduction of these artifacts is important for any quantitative
assessment of the tomographic images.
Advisor: Ishwar K. Sethi.
790 Directed Study - Message Passing Interface (MPI).
3 Credits.
MPI is
an attempt to develop a standard for the message passing parallel
computing paradigm that is portable and thus available on a large
number of platforms.
A study of MPI and its applicability to large scale
3D tomographic reconstruction problems will be investigated.
The recent text on MPI:
Using MPI, portable parallel programming with the message
passing interface by
William Gropp, Ewing Lusk, and Anthony Skjellum
was used as a reference.
Premilinary reports regarding MPI indicate
is is a very good tool for implementing parallel
tomographic reconstruction.
The portability of this library alllows execellent support for the
hardware comparisons expected to be performed as a part of my dissertation.
Advisor: Ishwar K. Sethi.
826 Seminar in Distributed Systems.
3 Credits.
Discussion of current research in the field.
Text: Distributed Systems, 2nd Edition, Edited by Sape Mullender,
Addison-Wesley/ACM Press, New York, 1993.
Instructor: Satyendra P. Rana.
886 Seminar Topics in Computer Vision and Pattern Recognition.
3 Credits.
Discussion of current research in the field.
Text: Computer Vision, Edited by Rangachar Kasturi and Ramesh C. Jain,
IEEE Computer Society Press, Los Alamitos, California, 1991.
Instructor: Jia-Guu Leu.
890 Graduate Seminar.
1 Credit.
Discussion of current research by the faculty and visitors.
Instructor: Narendra Goel.
999 Doctoral Dissertation Research and Direction.
20 Credits. 30 Credits Required.
Advisor: Ishwar K. Sethi.