Computer Science Colloquium

This talk describes a set of methods that I developed for nonrigid motion analysis, and their applications to some of my current projects in medical diagnostics, intuitive data exploration and molecular dynamics. The first part of the talk concentrates on recovery of observable (from images) and non-observable nonrigid motion descriptors. This is a very significant outcome for medical diagnostics applications since it provides the insight often required by physicians. Some important outcomes of estimating non-observable motion parameters include burn scar detection and tissue classification by recovered properties (burn scar estimation project), localizing unsafe modes of human motion or operation (Repetitive Stress Injury analysis project). Important theoretical contributions include algorithms for recovery of strains, initially unknown local geometry and material properties, point correspondences and 3-D structure under elastic/articulated motion assumption.

Next, I present a development of a real-time gesture-tracking system in support of intuitive visualization navigation. The recent trend in data growth is amplified by increasing requirements for interactive data access, display, exploration, analysis and collaboration. Goals of the system include supplying data manipulation parameters to interactive data exploration and collaborative visualization software, and to virtual reality systems. The talk describes a method performing range processing only when necessary and where necessary. Range data is processed only for non-static regions of interest. This is accomplished by a set of filters on the color, motion, and range data. A significant speedup is achieved.

Another recent research direction represents a quantitative analysis for a discovery in molecular dynamics. Recent simulations have shown that velocities of crack propagations in crystals under certain conditions can become supersonic, which is contrary to classical physics. In this part of the talk, I present a framework for tracking and motion analysis of crack propagations in crystals. This tracking is completely automated. Results supporting physical observations are presented in terms of both feature tracking and velocity analysis. Experimental results demonstrate the success of the described algorithms.

Leonid V. Tsap received a B.S. degree in Computer Science from the Kiev Civil Engineering Institute, Ukraine, in 1991, and his M.S. and Ph.D. degrees in Computer Science from the University of South Florida, Tampa, in 1995 and 1999, respectively. He also held full-time professional positions ranging from Computer Programmer to Senior Analyst and has taught a number of classes at USF. Leonid is a three-time winner of the annual University of South Florida USPS Scholarship Award, and a recipient of the Provost's Commendation for Outstanding Teaching by a Graduate Student. He also received University of South Florida Graduate Council's Outstanding Dissertation Prize. He is currently with the Center for Applied Scientific Computing at the University of California Lawrence Livermore National Laboratory.

Leonid is a member of the IEEE Computer Society and ACM. He is a member of the Editorial Board of the Pattern Recognition journal. His current research interests include image analysis/ computer vision, image synthesis and visualization, pattern recognition, perceptual user interfaces, physically-based modeling, and biomechanics. More specifically, Leonid is interested in nonrigid motion analysis and its applications to medical diagnostics, intuitive data exploration and molecular dynamics. His recent research resulted in 21 refereed publications. More information can be obtained from http://marathon.csee.usf.edu/~tsap and http://www.llnl.gov/CASC/people/tsap.