Application of extended Kalman filtering on aircraft pose prediction of image sequences

Abstract

In allusion to the character of monocular image sequences, a method based on extended Kalman filtering to predict the aircraft pose of image sequences is proposed. With α — β — γ stable state filtering technique, a mathematics model is built to realize the prediction of aircraft pose of image sequences. In the model, not only the influence of noise during the image process is considered, but also the shortcoming of low precision in the constant velocity model is overcomed. The derivation of acceleration is considered as white noise. The predictive curve plotted with Matlab proves that the maximum of error of using this method is about 3′. So its precision is higher and error standard deviation is lower than those of the constant velocity model.

References

[1] ZhangH., HuangZ.-H.. Journal of Optoelectronics • Laser, 2005, 16: 102

[2] ZhangZ.-X., SuG.-Z., ZhangJ.-Q.. Geomatics and Information Science of Wuhan University, 2004, 29: 287

[3] LiuY., WangY.-T., HuX.-M.. Computer Mea surement & Control, 2002, 10: 363

[4] J.S. Goddard and M.A. Abidi, SPIE, 3133, 189.

[5] Negenborn R, http://www.negenborn.net/kal-loc/thesis.pdf. 03/08/2005

[6] YuX.-F., DuanH.-B., GongH.-J.. Electronics optics and control., 2004, 11: 29

[7] XuL.-J., ChenY.-Z., CuiP.-Y.. Computer Measurement & Control., 2004, 12: 660