Daniel Freedma

Assistant Professor of Computer Science

Rensselaer Polytechnic Institute

freedman@cs.rpi.edu

A new approach to tracking using geometric active contours is presented. The class of objects to be tracked is assumed to be characterized by a probability distribution over some variable, such as intensity, colour, or texture. The goal of the algorithm is to find the region within the current image, such that the sample distribution of the interior of the region most closely matches the model distribution. Several criteria for matching distributions are examined, and the curve evolution equations are derived in each case. A particular flow is shown to perform well in two experiments.