Speaker: Hemant Tagare
Title: A Geometric Theory of Non-Rigid Registration
Affiliation: Yale University
Date: Friday, November 21, 2008
Place and Time: Room 101, Love Building, 3:35-4:30 pm
Refreshments: Room 204, Love Building, 3:00 pm

Abstract. Finding a non-rigid registration (or "warping") between images is a fundamental problem in bio-medical image processing. This talk presents a theory in which registration problems can be put in a unifying geometric framework. A key aspect of the theory is to treat the registration as geometric object (a manifold). This point of view shows very clearly how properties of registration algorithms are related to differential forms on the manifold and certain natural projection operators. Mathematical results of analyzing registration using this geometric formulation will be presented. The results will be applied to non-rigid registration of brain MRI images.

Biography. Hemant Tagare received his Ph.D. from Rice University in 1990 in Electrical Engineering. Since then he has been at Yale University, initially as a postdoc and now as tenured Associate Professor. His research interests are in bio-medical image processing. He develops algorithms and mathematical theories of image processing. His current research interests are in segmentation, registration, and shape analysis.