A Face Tracking Method Based on Camshift and SCKF in Rapid Moving Process

Abstract:

Aiming at the problem of face tracking under rapid moving process, a fast and robust tracking method is proposed. The possible position of face detected by the Camshift algorithm in the next frame is predicted by the square-root cubature Kalman filte (SCKF). Then, the localization and tracking of face are got frames by frames. The experimental results show that: the use of SCKF to solve the nonlinear effect caused by non-uniform motion of face and overcome the target loss problem of the linear Kalman algorithm. The proposed method greatly improves the tracking accuracy of face in the process of rapid movement.

• Periodical: Applied Mechanics and Materials (Volumes 668-669)
• Pages: 1025-1028
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.668-669.1025
• Online since: October 2014
• Authors: Fu Cheng Cao*, Xiao Xue Xing
• Keywords: Camshift, Computer Vision, Cubature Kalman Filter, Face Tracking

References

[1] Yang G Z, Huang T S. in: Human face detection in a complex background, Pattern Recognition, 27(1), (1994).

[2] Moghaddam B, Pentland A. in: Probabilistic visual learning for object representation, IEEE Trans. PA MI, 19(7), (1997).

[3] Rowley H A, Baluja S, Kanade T. in: Neural network-based face detection, IEEE Trans. PAMI, 20(1), (1998).

[4] Sung K, Poggio T. in: Example-based learning for view-based human face detection, IEEE Trans. PA MI, 20(1), (1998).

[5] Osuna E, Freund R, Girosi F. in: Training support vector machines: an application to face detection, In Proc. of CVPR, Puerto Rico, (1997).

[6] Chu Hongxia, Ye Shujiang, et al. in: Object tracking algorithm based on Camshift algorithm combinating with difference in frame, IEEE Conf. on Automation and Logistic, (2007).

DOI: 10.1109/ical.2007.4338529

[7] Nummiaro K, Koller-Meier E, L Van Gool. in: Color features for tracking non- rigid objects, Special Issue on Visual Surveillance, Chinese Journal of Automation, 29(3), (2003).

[8] HU Zhangfang, et al. in: The research of face tracking algorithm based on hybrid Kalman and Camshift filter, Journal of Chongqing University of Arts and Sciences, 32(3), (2013).

[9] TIAN Gang, HU Ruimin, et al. in: Object tracking algorithm based on meanshift algorithm combining with motion vector analysis, 2009 First International Workshop on ETCS, (2009).

DOI: 10.1109/etcs.2009.225

[10] Arasaratnam, S. Haykin, T. R. Hurd. In: Cubature Kalman filtering for continuous-discrete systems: Theory and simulations, IEEE Trans. on Signal Processing, 58(10), (2010).

DOI: 10.1109/tsp.2010.2056923

[11] http: /www. cems. uvm. edu/~gmirchan/classes/EE275/2012/Projects/FaceDetection.