Real-time tracking of non-rigid objects using mean shift

Abstract
A new method for real-time tracking of non-rigid objectsseen from a moving cameraisproposed. The centralcomputational module is based on the mean shiftiterations and finds the most probable target position inthe current frame. The dissimilarity between the targetmodel (its color distribution) and the target candidatesis expressed by a metric derivedfrom the Bhattacharyyacoefficient. The theoretical analysis of the approachshows that it relates to the Bayesian framework whileproviding a ...

This publication has 20 references indexed in Scilit: