飞行器飞行目标定位组合导航精度预测仿真
|
摘要
在飞行器飞行目标定位组合导航系统中,高精度滤波算法对飞行器目标定位导航解算起着至关重要的作用。为了提高飞行器组合导航系统滤波算法的精度,针对粒子滤波(PF)算法中的粒子退化问题,提出了支持向量机(SVM)和容积卡尔曼滤波(CKF)辅助粒子滤波的SVM-CPF算法。算法通过采用CKF来生成粒子滤波的重要性密度函数,并利用SVM重采样获得多样性的粒子,使滤波性能明显改善,提高飞行器飞行目标定位组合导航精度。将上述算法应用在飞行目标定位捷联惯导(SINS)/全球定位系统(GPS)组合导航系统中,仿真结果表明,改进的滤波算法能提高飞行器目标定位精度,系统优化性能明显优于粒子滤波(PF)以及容积卡尔曼滤波(CKF)。
The traditional semi-global stereo matching algorithm well balances the accuracy and real-time performance of matching algorithm, however, which has a high mismatching rate of the depth discontinuity regions in the image processing. In view of the above problem, the unmanned aerial vehicle(UAV) Semi-global Matching Algorithm through Fusion of Fast Mean Shift(FMS)Image Segmentation is proposed. The FMS image segmentation algorithm was integrated into the global energy function of the matching algorithm to reconstruct a more reasonable global energy function, which solves the problem of assigning the same penalty coefficient for disparity mutations due to different reasons. The simulation test based on the algorithm was carried out on the Middle burry test platform. And based on the proposed algorithm, a real scene image filmed by UAV experiment was conducted. The results show that the algorithm effectively reduces the mismatching rate of the images in the depth disparity mutation regions and balances the matching accuracy and matching speed of the algorithm. Therefore, the improved algorithm can meet the requirements of UAV vision assisted navigation.
The traditional semi-global stereo matching algorithm well balances the accuracy and real-time performance of matching algorithm, however, which has a high mismatching rate of the depth discontinuity regions in the image processing. In view of the above problem, the unmanned aerial vehicle(UAV) Semi-global Matching Algorithm through Fusion of Fast Mean Shift(FMS)Image Segmentation is proposed. The FMS image segmentation algorithm was integrated into the global energy function of the matching algorithm to reconstruct a more reasonable global energy function, which solves the problem of assigning the same penalty coefficient for disparity mutations due to different reasons. The simulation test based on the algorithm was carried out on the Middle burry test platform. And based on the proposed algorithm, a real scene image filmed by UAV experiment was conducted. The results show that the algorithm effectively reduces the mismatching rate of the images in the depth disparity mutation regions and balances the matching accuracy and matching speed of the algorithm. Therefore, the improved algorithm can meet the requirements of UAV vision assisted navigation.
引文
[1] O Cappe, S J Godsill, E moulines. An overview of existing methods and recent advances in sequential Monte Carlo[J]. IEEE Proceedings, 2007,95(5):899-924.
[2] N Bouaynaya, W Qu, D Schfeld. An on line motion-based particle filter for head tracking applications[C]. IEEE International Conference on Acoustics, Speech, and Signal Processing, Philadelphia, 2005:225-228.
[3] J V Candy. Bootstrap particle filter[J]. IEEE Signal Processing Magazine, 2007,24(4): 73-85.
[4] M S Arulampalam, etal. A tutorial on particle filters for online nonlinear/non-gaussian Bayesian tracking[M]. Signal Processing, 2002,50(2):174-188.
[5] A Mohammadi, A Asif. Consesus-based distributed unscented particle filter[C]. IEEE Statistical Signal Processing Workshop(SSP). Nice, 2011:237-240.
[6] 常天庆,等. 一种改进重采样的粒子滤波算法[J]. 计算机应用研究, 2013,30(3):748-750.
[7] 刘建业,等. 导航系统理论与应用[M]. 西安:西北工业大学出版社, 2010:350-354.
[8] I Arsaratnam, S Haykin. Cubature Kalman Filters[J]. Automatic Control, 2009,54(6): 1254-1269.
[9] V N Vapnik. The nature of statical learning.Berlin[M]. Springer, 1995.
[10] V N Vapnik. Support vector method for multivariate density estimation[C]. Advances in Neural information Processing System. Cambrige: M It Press, 2000.
[11] J Weston, et al. Support vector density estimation[C]. Advances in Kernel Methods-SV Learning. Cambridge: M It Press, 1999.
[2] N Bouaynaya, W Qu, D Schfeld. An on line motion-based particle filter for head tracking applications[C]. IEEE International Conference on Acoustics, Speech, and Signal Processing, Philadelphia, 2005:225-228.
[3] J V Candy. Bootstrap particle filter[J]. IEEE Signal Processing Magazine, 2007,24(4): 73-85.
[4] M S Arulampalam, etal. A tutorial on particle filters for online nonlinear/non-gaussian Bayesian tracking[M]. Signal Processing, 2002,50(2):174-188.
[5] A Mohammadi, A Asif. Consesus-based distributed unscented particle filter[C]. IEEE Statistical Signal Processing Workshop(SSP). Nice, 2011:237-240.
[6] 常天庆,等. 一种改进重采样的粒子滤波算法[J]. 计算机应用研究, 2013,30(3):748-750.
[7] 刘建业,等. 导航系统理论与应用[M]. 西安:西北工业大学出版社, 2010:350-354.
[8] I Arsaratnam, S Haykin. Cubature Kalman Filters[J]. Automatic Control, 2009,54(6): 1254-1269.
[9] V N Vapnik. The nature of statical learning.Berlin[M]. Springer, 1995.
[10] V N Vapnik. Support vector method for multivariate density estimation[C]. Advances in Neural information Processing System. Cambrige: M It Press, 2000.
[11] J Weston, et al. Support vector density estimation[C]. Advances in Kernel Methods-SV Learning. Cambridge: M It Press, 1999.