Probabilistic Image-Based Tracking: Improving Particle Filtering
Condensation is a widely-used tracking algorithm based on particle filters. Although some results have been achieved, it has several unpleasant behaviours. In this paper, we highlight these misbehaviours and propose two improvements. A new weight assignment, which avoids sample impoverishment, is presented. Subsequently, the prediction process is enhanced. The proposal has been successfully tested using synthetic data, which reproduces some of the main difficulties a tracker must deal with.
KeywordsParticle Filter Visual Tracking Background Clutter Weight Assignment Recursive Estimation
Unable to display preview. Download preview PDF.
- 2.Doucet, A.: On Sequential Simulation-Based Methods for Bayesian Filtering. CUED/F-INFENG/TR 310. University of Cambridge (1998) Google Scholar
- 4.King, O., Forsyth, D.A.: How Does Condensation Behave with a Finite Number of Samples? In: ECCV proceedings, vol. (1), pp. 695–709 (2000)Google Scholar
- 5.Russell, R., Norvig, P.: Artificial Intelligence, a Modern Approach, ch. 13-15. Prentice-Hall, Englewood Cliffs (2003)Google Scholar
- 6.van der Merwe, R., Doucet, A., de Freitas, N., Wan, E.: The Unscented Particle Filter. CUED/F-INFENG/TR 380. University of Cambridge (2000) Google Scholar
- 7.Varona, X., Gonzàlez, J., Roca, X., Villanueva, J.J.: iTrack: Image-based Probabilistic Tracking of People. ICPR (3), 7122–7125 (2000)Google Scholar