A combined object-tracking algorithm for omni-directional vision-based AGV navigation

Abstract

A combined object-tracking algorithm that realizes the realtime tracking of the selected object through the omni-directional vision with a fisheye lens is presented. The new method combines the modified continuously adaptive mean shift algorithm with the Kalman filter method. With the proposed method, the object-tracking problem when the object reappears after being sheltered completely or moving out of the field of view is solved. The experimental results perform well, and the algorithm proposed here improves the robustness and accuracy of the tracking in the omni-directional vision.

Keywords

Kalman Filter / Color Histogram / Search Window / Fisheye Lens / Shift Algorithm

References

[1] WANG Yongzhong, Fish-Eye Lens Optics, Science Press, 2006. (in Chinese)

[2] JiaY., LvH., XuA., LiuW.. Chinese J. Computers, 2000, 23: 1216

[3] ComaniciuD., RameshV., MeerP.. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2000, 2: 142

[4] SunK., LiuS.. Information and Control, 2009, 38: 9

[5] KangW., DingX., LiuG., AoL.. Journal of Optoelectronics Laser, 2008, 19: 135

[6] Donghe Yang and Jinaong Xia, Intelligent Systems and Applications, 2009.

[7] WeiK., ZhaoY., PanQ.. Journal of Optoelectronics Laser, 2008, 19: 213

[8] Stefano Messelodi, Carla Maril Modena, Nicola Segata and Michele Zanin, ICIAP 2005, Lecture Notes in Computer Science 3617, 163 (2005).

[9] S. M. Bozic, Digital and Kalman Filtering, Science Press, 174 (1984). (in Chinese)

[10] HouZ.-Q., HanC.-Z.. Acta Automatica Sinica, 2006, 32: 603

[11] Greg Welch and Gray Bishop, An Introduction to the Kalman Filter, UNC-Chapel Hill, TR 95-014 (2004).