Seminars/Colloquia/Invited Talks

Seminars

Yuri Boykov

Optimization of Surface Functionals Using Discrete Graph Cut Algorithms

PLACE:	Clark 110
EVENT:	CIS Seminar
DATE:	March 11, 2008
TIME:	1:00 - 2:00 PM

Abstract

Regular grid graphs and complexes can be used to approximate geometric surface functionals. Standard combinatorial algorithms for max-flow/min-cut problems such as augmenting paths, push-relabel, pseudo-flow, and parametric max-flow techniques can efficiently compute (in low-order polynomial time) global optima solutions for the corresponding approximations of geometric problems widely used in computer vision, graphics, and medical imaging. We review a range of applicationsand discuss some computational issues (what geometric functionals can be approximated on graphs,metrication errors, global optimization vs. local optimization, running-time and memory efficiency).

Brief Biography:

Dr. Yuri Boykov received his "Diploma of High Education" with honors at Moscow Institute of Physics and Technology in 1992 and completed his Ph.D. at the department of Operations Research at Cornell University, NY, in 1996. He first became interested in combinatorial approach to generic energy minimization problems in low-level vision while he was a post-doc at the CS department at Cornell University, NY. Jointly with Olga Veksler and Ramin Zabih, Yuri developed a powerful "alpha-expansion" aproximation method for estimating piece-wise smooth and piece-wise constant Markov Random Fields models which are currently widely used in computer vision community. Later, as a scientist at Siemens Research Institute in Princeton, he developed graph-cuts methodology for object extraction from ND imagery. Such global optima hypersurface extraction methods became a benchmark standard in medical applications and in many 3D problems in computer vision. Currently, Yuri is an Assistant Professor at the department of Computer Science at the University of Western Ontario, Canada. He recently receive a Florence Bucke Sience award from the faculty of Science. He is interested in problems of segmentation, restoration, registration, stereo, surface fitting, feature-based object recognition, tracking, photo-video editing, learning graph-based representation models, computational geometry, and others.