An EM-like algorithm for color-histogram-based object tracking

Author

Zivkovic, Zoran ; Krose, Ben

Author_Institution

Intelligent & Autonomous Syst. Group, Amsterdam Univ., Netherlands

Volume

1

fYear

2004

fDate

27 June-2 July 2004

Abstract

The iterative procedure called ´mean-shift´ is a simple robust method for finding the position of a local mode (local maximum) of a kernel-based estimate of a density function. A new robust algorithm is given here that presents a natural extension of the ´mean-shift´ procedure. The new algorithm simultaneously estimates the position of the local mode and the covariance matrix that describes the approximate shape of the local mode. We apply the new method to develop a new 5-degrees of freedom (DOF) color histogram based non-rigid object tracking algorithm.

Keywords

covariance matrices; estimation theory; image colour analysis; iterative methods; optimisation; tracking; 5 DOF; 5-degrees of freedom; EM algorithm; color histogram; covariance matrix; iterative procedure; kernel based density function estimation; local mode; mean shift procedure; nonrigid object tracking algorithm; position estimation; robust algorithm; Covariance matrix; Density functional theory; Histograms; Intelligent systems; Iterative algorithms; Iterative methods; Maximum likelihood estimation; Probability density function; Robustness; Shape;

fLanguage

English

Publisher

ieee

Conference_Titel

Computer Vision and Pattern Recognition, 2004. CVPR 2004. Proceedings of the 2004 IEEE Computer Society Conference on

ISSN

1063-6919

Print_ISBN

0-7695-2158-4

Type

conf

DOI

10.1109/CVPR.2004.1315113

Filename

1315113