Infrared target tracking based on multi-feature fusion under motion platform

Abstract

In order to realize infrared target tracking accurately under motion platform, and make up for the shortcoming of the nuclear density estimation based on gradation feature, an adaptive kalman-mean shift algorithm based on multi-feature fusion is proposed. The target model based on edge-gradation feature fusion is applied in the mean shift algorithm. The starting position at present of an infrared target is predicted by a kalman filter, and then a scale updating item of tracking window is appended based on the relationship between mutual information and the object scale. Then the moving object, especially the object with a variable scale, is adaptively tracked under motion platform. Experimental results demonstrate that the adaptability of mean shift algorithm is enhanced by the improved scheme, which can be applied in the process of long time tracking for the object with a variable scale.

Keywords

Mutual Information / Kalman Filter / Target Model / Motion Platform / Edge Feature

References

[1] LiL., LiJ.-s., YeX.. Infrared and Laser Engineering, 2007, 36: 229

[2] ComaniciuD., RameshV., MeerP.. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003, 25: 564

[3] HuangF., ZhouX., ZhouS. d.. Infrared and Laser Engineering, 2005, 34: 324

[4] WangX.-h., WangJ., TianM.. Journal of Image and Graphics, 2008, 13: 586

[5] ZhangX.-g., ZhaoE.-l., WangY.-j.. Optical Technique, 2007, 33: 226

[6] ZhangX.-g., ZhangY., WangY.-j.. Opto-Electronic Engineering, 2008, 35: 45

[7] Fan You-fu, the Application of Mean Shift Algorithm on Object Tracking, Wuhan University of Science and Technology, 2007, 19. (in Chinese)

[8] QianH.-m., MaoY.-b., WangZ.-q.. Journal of Image and Graphics, 2007, 12: 245

[9] KangW.-j., DingX.-m., LiuG.-l.. Journal of Optoelectronics·Laser, 2008, 19: 135