Seminar Announcement
This seminar is sponsored by the IEEE Upper Mon subsection

Abstract: Although the problem of automatic fingerprint matching has been extensively studied, it is nevertheless, not a fully solved problem. There are a myriad of unresolved issues that need to be addressed in this rather popular biometric technique. In this talk I will describe some of these issues and techniques to solve them. The presentation will include discussions on the following:

1. To augment minutiae information with the underlying texture information a hybrid fingerprint representation scheme will be presented. The proposed hybrid system is shown to perform better than a purely minutiae-based fingerprint matching system.
2. A fingerprint mosaicking scheme to construct a composite fingerprint template using multiple impressions of a finger will be outlined. The method presented is useful to register and generate an elaborate fingerprint from multiple partial prints acquired using small-sized fingerprint sensors.
3. A non-linear distortion model to account for warping in fingerprints will be described. The warping model is seen to improve the matching performance of a fingerprint system.
4. Finally, techniques to combine fingerprint evidence of a user with other biometric traits (viz., face and hand geometry) will be presented. To enhance user convenience, a learning technique is employed to compute user-specific parameters for each biometric trait.

Speaker Bio: Arun Ross obtained the M.S. degree in Computer Science and Engineering from Michigan State University (USA) in 1999, and the B.E. (Hons.) degree in Computer Science from the Birla Institute of Technology and Science, Pilani (India), in 1996. He is currently a Ph.D. candidate at Michigan State University and his dissertation research is on information fusion in fingerprint authentication. His areas of interest include statistical pattern recognition, machine learning, computer vision, and biometric authentication.