Vision and Robotics Seminar
Department of Computer Science
University of Haifa
will speak on
Bayesian Anecdotes - Two Applications
of the Bayesian Method to Computer Vision
This talk will survey two applications of the maximum a-posteriori (MAP) principle, derived using Bayes' rule. The first is the problem of determining the ``style" of an image. An application to identifying an artist given a painting will be presented, which yields results consistent with human intuition. The second concerns motion recovery, for which we offer a new paradigm based on computing the MAP estimate of the motion parameters and FOE. The main novelty is that all domain-source point combinations are considered, as opposed to a single ``optimal" combination. While this involves the optimization of non-trivial cost functions, the results are considerably superior to those of the so-called algebraic and geometric methods, especially when there is a lot of noise present, and/or when the measurement points approach a degenerate configuration.
Joint work with Ilan Shimshoni, Liran Goshen, and Michael Werman.
The lecture will take place in the
Lecture Hall, Room 1, Ziskind Building
on Thursday, December 20, 2001