Associate Chair, Mathematics
Associate Professor
 |
438 Carver Hall Department of Mathematics Iowa State University Ames, IA 50011 Telephone: (515) 294-2941 Fax: (515) 294-5454 E-mail: davidson@iastate.edu Ph.D., University of Florida, 1989 M.S., University of Florida, 1986 B.A., Mount Holyoke College, 1979 |
Research interests
Image modeling and analysis, stochastic image modeling, steganalysis
Research Projects
- A Steganalyzer Package for Forensic Applications, Midwest Forensic Resource Center/NIJ, Ames Lab, 2009.
- An Artificial Neural Network for Wavelet Steganalysis. Midwest Forensic Resource Center, Ames Lab, Ames, Iowa, 2004-6.
- The Singular Value Decomposition for Steganography.
Publications
- J. Davidson, J. Jalan. Feature Selection for Steganalysis using the Mahalanobis Distance, SPIE Electronic Imaging, Media Forensics and Security XII, Proc. of SPIE Volume 7541, San Jose, CA, January 18-20, 2010.
- C. Bergman, J. Davidson. An Artificial Neural Network for Wavelet Steganalysis, Final Report, Midwest Forensic Resource Center, January, 2007. PDF
- J. Davidson, C. Bergman, E. Bartlett. An artificial neural network for wavelet steganalysis. Proceedings of SPIE, Vol. 5916, Mathematical Methods in Pattern and image Analysis, pp. 1-10, 2005. PDF
Current Research Description
Steganalysis is used to detect hidden content in innocuous images. Many successful steganalysis algorithms use a large number of features relative to the size of the training set and suffer from a "curse of dimensionality:" large number of feature values relative to training data size. High dimensionality of the feature space can reduce classification accuracy, obscure important features for classification, and increase computational complexity. Our research towards this end investigates a filter-type feature selection algorithm that selects reduced feature sets using the Mahalanobis distance measure, and develops classifiers from the sets. The experiment is applied to a well-known JPEG steganalyzer, and shows that using our approach, reduced-feature steganalyzers can be obtained that perform as well as the original steganalyzer. The steganalyzer is that of Pevny et al. (SPIE, 2007) that combines DCT-based feature values and calibrated Markov features. Five embedding algorithms are used. Our results demonstrate that as few as 10-60 features at various levels of embedding can be used to create a classifier that gives comparable results to the full suite of 274 features.
Correct classification of images with hidden payload requires sophisticated detection algorithms. It involves the use of pattern recognition techniques, statistical and mathematical analysis, efficient programming skills, knowledge of image processing algorithms, plus a myriad of other skills from the areas of computer science, electrical and computer engineering, detection and information theory, and stochastic processes. Our research team is investigating the use of stochastic spatial models to help model the effects of inserting the message bits into the image, particularly partially ordered Markov models. These models allow a structured and theoretical approach to be applied to steganalysis.
We are looking for a graduate student to fill a possible research assistantship opening in 2010. Please contact Dr. Jennifer Davidson for further inquiries.
Courses taught
- Spring semesters: Steganalysis, Steganography and Watermarking.
- Fall semesters: Calculus I, Calculus II
Current Students
- Jaikishan Jalan (MS, Computer Science): Blind JPEG Steganalyzer
- Faryal Awan (MS, Electrical and Computer Engineering)
