A Robust Tracking Algorithm in Crowded Environment

Abstract:

Reliable tracking of multiple people in cluttered or complex situations is a challenging visual surveillance problem since the high density of objects results in occlusion. In order to deal with this problem, multiple synchronized cameras were mounted at various heights in our experiment. To ensure the existence of the homography, it is necessary to assume that different views share a common dominant ground plane. Thus, corresponding people have to be located within the multi-camera surveillance system, accurate multi-people localizing is an important prerequisite to reliable tracking. In this paper, we present a novel approach to fusing foreground information on planes of different height from multiple views to increase accurateness of localization. Our method does not require fully camera calibration. First we obtain the foreground likelihood maps in each view by modeling the background using codebook algorithm. Then we compute homographies induced by multiple planes and obtain the localization of multiple people at multiple planes. Finally, we adopt an algorithm of feature-based using Kalman filter motion to handle multiple objects tracking at multiple plane. The experimental results show that our method is valid and has nice robustness to the occlusion in crowed environments.

• Periodical: Applied Mechanics and Materials (Volumes 157-158)
• Pages: 1336-1339
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.157-158.1336
• Online since: February 2012
• Authors: Ming Xin Jiang, Hong Yu Wang
• Keywords: Homography, Localizing, Multi-Camera, Multiple Planes

References

[1] J. Kang, I. Cohen, and G. Medioni, Continuous Tracking within and across Camera Streams, Proceedings of the IEEE International Conference on Computer Vision and Pattern Recognition, pp.267-272, June (2003).

DOI: 10.1109/cvpr.2003.1211363

[2] M X Jiang, M Li, H Y Wang. A Robust Combined Algorithim of Object Tracking Based on Moving Object Detection,. Proceedings of the International Conference on Intelligent Control and Information Processing. Dalian, China: IEEE, pp.619-622, August (2010).

DOI: 10.1109/icicip.2010.5564172

[3] T. Zhao, R. Nevatia, Tracking multiple humans in crowded environment, Proceedings of the IEEE International Conference on Computer Vision and Pattern Recognition, pp.459-466, January (2004).

DOI: 10.1109/cvpr.2004.1315192

[4] J. Kang, I. Cohen, G. Medioni, Multi-Views tracking within and across uncalibrated camera streams, First ACM SIGMM international workshop on Video surveillance, Berkeley, California, pp.21-33, November (2003).

DOI: 10.1145/982452.982456

[5] D. Comaniciu, V. Ramesh, and P. Meer, Kernel based object tracking, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 25, p.564–577, (2003).

DOI: 10.1109/tpami.2003.1195991

[6] K. Kim, T.H. Chalidabhongse, D. Harwood, and L.S. Davis, Real-Time foreground-background segmentation using codebook model, Real-Time Imaging, vol. 11, pp.167-256, (2005).

DOI: 10.1016/j.rti.2004.12.004

[7] C. Rother, A new approach for vanishing point detection in architectural environments, Proceedings of the 13th British Machine Vision Conference, (2002).

[8] F. Lv, T. Zhao and R. Nevatia, Camera calibration from video of a walking human, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 28, pp.1513-1518, (2006).

DOI: 10.1109/tpami.2006.178