Decision, Control and Optimization Seminar

Decision and Control Laboratory, Coordinated Science Laboratory

Constructing Robust Topological and Functional Image Descriptors

Professor Edgar Lobaton

North Carolina State University, Dept. of ECE

Wednesday, April 18, 2012

3:00 PM to 4:00 PM

B02 CSL

Abstract
      Local image descriptors are a fundamental block for several computer vision applications. These descriptors must be robust to a variety of nuisance operators, some of which may include deformations, noise and image discretization. Traditionally, features assume an affine deformation model which is consistent with change of perspective and rigid body motion. However, applications associated with tracking of objects that undergo more generic deformations (e.g., biological tissue during a surgical procedure, cloth and human tracking) require robust descriptors that are invariant to generic deformations. We introduce a methodology for constructing local functional and topological descriptors that are invariant to bounded deformations and robust to image discretization. This approach does not require any sampling of the parameter space for it to be robust and no distributional assumptions are made. Furthermore, this methodology guarantees the identification of matching feature points using a hierarchical approach which guarantees that few possible matching points are lost in the process. This is accomplished via the construction of a cover for any possible matching point and its subsequent refinement.

Biography
Dr. Lobaton received the B.S. degree in mathematics and the B.S. degree in electrical engineering from Seattle University in 2004. He completed his Ph.D. in electrical engineering and computer sciences from the University of California, Berkeley in 2009.

He is currently an Assistant Professor in the Department of Electrical and Computer Engineering at North Carolina State University. His areas of research include robotics, sensor networks and computer vision. He works on applications ranging from surveillance using smart camera systems to motion planning for medical robotic applications. Prior to joining NCSU, he was awarded the 2009 Computer Innovation Fellows post-doctoral fellowship award and conducted research in the Department of Computer Science at the University of North Carolina at Chapel Hill. He was also engaged in research at Alcatel-Lucent Bell Labs in 2005 and 2009.

PLEASE JOIN US FOR COOKIES AND COFFEE PRIOR TO THE SEMINAR

IN ROOM 154 CSL AT 2:40PM