Title:
The CONDENSATION algorithm for object-tracking and more

Eric Spellman, CISE

Abstract:
The CONDENSATION algorithm, developed by A. Blake and M. Isard, was intended
for tracking an object through a video sequence.  However, the algorithm's
framework is quite general and could be extended to other applications.
First we will examine the framework of the algorithm and how it is applied
to image-tracking.  This framework casts the tracking problem in
probabilistic terms by representing and propogating the distribution over
the possible states of the tracked object.  (This includes its position,
orientation, etc.)  Then we will consider applying the algorithm to other
problems.  Here we will analyze how CONDENSATION's probabilistic nature
forms the algorithm's search strategy, and we will identify the type of
problems where such a search strategy might be beneficial.  (Audience
participation is strongly encouraged here.)