Title :
Variable Number of "Informative" Particles for Object Tracking
Author :
Huang, Yu ; Llach, Joan
Author_Institution :
Thomson Corp. Res., Princeton
Abstract :
Particle filter is a sequential Monte Carlo method for object tracking in a recursive Bayesian filtering framework. The efficiency and accuracy of the particle filter depends on two key factors: how many particles are used and how these particles are re-located. In this paper, we estimate the number of required particles using the Kullback-Leibler distance (KLD), which is called KLD-sampling, and we use a hybrid dynamic model to generate diversified particles, which suits object\´s agile motion. Besides, we employ the mean shift analysis as a local mode seeking mechanism to make each particle more "informative". We demonstrate the performance of the proposed algorithm tracking the ball in sports video clips.
Keywords :
Bayes methods; Monte Carlo methods; particle filtering (numerical methods); target tracking; Kullback-Leibler distance; informative particles; object tracking; particle filter; recursive Bayesian filtering; sequential Monte Carlo method; Bayesian methods; Computer vision; Filtering; Hybrid power systems; Motion estimation; Particle filters; Particle tracking; Robustness; State estimation; Video compression; Sampling methods; Tracking;
Conference_Titel :
Multimedia and Expo, 2007 IEEE International Conference on
Conference_Location :
Beijing
Print_ISBN :
1-4244-1016-9
Electronic_ISBN :
1-4244-1017-7
DOI :
10.1109/ICME.2007.4285053
