Todd Wittman

Assistant Professor
Department of Mathematics
The Citadel
HOME
TEACHING
RESEARCH
STUDENTS


My research focuses on the application of mathematics to problems in image processing. Mathematical image processing involves many different areas of applied math, including PDEs, the calculus of variations, numerical analysis, wavelets, and machine learning.


Current Projects
Medical Imaging

Doctors use a variety of imaging devices (MRI, CT, PET, etc.), but the images are often noisy and difficult to read. (Hence the need for trained radiologists.) Image processing can assist doctors in detecting and diagnosing medical problems.


Geospatial Imaging
in collaboration with Prof. Andrea Bertozzi (UCLA) and Prof. Stanley Osher (UCLA)

Under government sponsorship, our research group is exploring the application of PDE and variational image processing methods to aerial and satellite images.


Past Projects
Variational Methods for Image Super-Resolution
Academic Advisor: Prof. Fadil Santosa (University of Minnesota)
Fall 2003 - Summer 2006

As part of my PhD thesis research, I explored the application of variational models to image zooming and super-resolution.


MANI: Manifold Learning Toolkit
Course Instructor: Prof. Gilad Lerman (University of Minnesota)
Spring 2005

As part of a graduate course on geometric data analysis, I wrote a Matlab GUI that compared 8 different Non-Lineard Dimension Reduction (NLDR) methods. The toolkit has been used by researchers for exploring manifold learning and screenshots of the GUI have appeared in several research articles.


Barcode Image Processing
Academic Advisor: Prof. Fadil Santosa (University of Minnesota)
Industrial Advisor: Dr. Miroslav Trajkovic (Symbol Technologies)
Summer 2004

Working as a research intern for Symbol Technologies, the world's largest manufacturer of hand-held barcode scanners, I developed techniques for decreasing blur and improving resolution in barcode imaging.


Data Mining the Stock Market
Course Instructor: Prof. Vipin Kumar (University of Minnesota)
Fall 2002

For a final project in a graduate course on data mining, I looked for correlations between different industries based just on their stock histories. Some observed patterns seem natural, such as internet and software stock prices being strongly correlated. Other patterns were surprising, such as a rise in pharmaceutical stocks frequently coinciding with a decline in automotive stocks.


Pattern Recognition in Railroad Maintenance
Academic Advisor: Prof. Fadil Santosa (University of Minnesota)
Industrial Advisor: Kenneth Iverson (Loram Maintenance of Way Inc.)
Summer 1997 - Summer 1999

My work for my Master's Thesis arose from an industrial internship at Loram, a railroad maintenance firm based in Hamel, MN. My project was to find the optimal combination of grinding patterns to smooth a rail into an ideal mushroom shape.


Todd Wittman
Department of Mathematics
The Citadel
E-mail: twittman @ citadel.edu