Adaptive Fusion Color and Haar-like Feature Object Tracking Based on Particle Filter

DOI

10.2991/icmt-13.2013.190How to use a DOI?

Keywords

Particle filter· observation model· haar-like· confidence

Abstract

The conventional kernel color histogram based particle filter (CHGPF) object tracking algorithm suffers from cluttered backgrounds and illumination changes and results in tracking lost easily. In this paper, a particle filter based object tracking algorithm fusing color and haar-like feature (CHAPF) is proposed. As haar-like features descript the object structure in multi-scale model, the haar-like feature based Semi-Supervised On-line Boosting (HSEMOB) tracking algorithm can well distinguish between the object and background, and be adaptive to changes of the object and environment by updating haar-like features. In our proposed algorithm, the results of HSEMOB and CHGPF are weighted adaptively to obtain a new particle filter weights. So, the color-based and haar-like based observations are fused into the framework of particle filter to make it adapt to the change of scene. Algorithm performance is tested with the environment complex videos. Experimental results show that the algorithm has better accuracy compared to CHGPF and HSEMOBPF for the scene of an object has similar appearance to the background or changes of illumination.

Copyright

© 2013, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Xian-Bing Ma
AU  - Shui-Fa Sun
AU  - Yin-Shi Qin
AU  - Song Hu
AU  - Bang-Jun Lei
PY  - 2013/11
DA  - 2013/11
TI  - Adaptive Fusion Color and Haar-like Feature Object Tracking Based on Particle Filter
BT  - Proceedings of 3rd International Conference on Multimedia Technology(ICMT-13)
PB  - Atlantis Press
SP  - 1556
EP  - 1563
SN  - 1951-6851
UR  - https://doi.org/10.2991/icmt-13.2013.190
DO  - 10.2991/icmt-13.2013.190
ID  - Ma2013/11
ER  -