Observer-Theoretic Methods for Computer Vision
Dr. Patricio VelaAssistant Professor, Electrical and Computer Engineering, Georgia Tech
Date: Wednesday, April 2, 2008 at 12:00 PM
Room: ELAB 21
Abstract
Many computer vision algorithms utilize a Bayesian assumption together with a Markov update model in order to obtain tractable derivations that are also algorithmically viable. Given that the ultimate task of many computer vision processes is to estimate the visual state of a system from image observations, we propose to define and derive computer vision algorithms as observers. More specifically the talk will focus on efforts to define observers for target tracking algorithms. A comparison and discussion of the relation to existing methods will conclude the talk.
Presenter Bio
Patricio Vela is an Assistant Professor in the School of ECE at Georgia Tech. His current research interests are in the benefits that a control-theoretic interpretation of computer vision algorithm can provide to the fidelity and principled derivation of said algorithms.
