计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (36): 129-133.

• 数据库、信号与信息处理 • 上一篇    下一篇

卡尔曼滤波与粒子滤波之间跟踪模式的优化

张  萌1,陈  恳1,李  娜2,惠  明1   

  1. 1.宁波大学 信息科学与工程学院,浙江 宁波 315211
    2.中国科学院 电子学研究所,北京 100190
  • 出版日期:2012-12-21 发布日期:2012-12-21

Optimization of tracking mode between Kalman Filter and Particle Filter

ZHANG Meng1, CHEN Ken1, LI Na2, HUI Ming1   

  1. 1.College of Information Science and Engineering, Ningbo University, Ningbo, Zhejiang 315211, China
    2.Institute of Electronics, Chinese Academy of Sciences, Beijing 100190, China
  • Online:2012-12-21 Published:2012-12-21

摘要: 鉴于卡尔曼滤波(Kalman Filter,KF)和粒子滤波(Particle Filter,PF)都是贝叶斯估计的一种,粒子滤波比卡尔曼滤波应用广泛,而卡尔曼滤波比粒子滤波使用简便,提出了一种算法在卡尔曼滤波和粒子滤波之间切换的跟踪模式。定义出算法性能评价参数,使参数可以在线反映算法的好坏;通过仿真使噪声不满足卡尔曼使用条件,确定切换是否可行,结合实际情况定义切换条件;应用至实际视频中。结果证明,卡尔曼滤波与粒子滤波间跟踪模式的优化是可行的。

关键词: 新的跟踪模式, 卡尔曼滤波, 粒子滤波, 算法性能评价, 算法切换

Abstract: Based on Kalman Filter and Particle Filter are all kinds of Bayesian estimation, the PF uses wider than KF, while the KF is easier than PF, proposing a new tracking mode that the tracking algorithms can be switched between KF and PF. This paper defines the parameter which can evaluate the algorithms’ performance online, makes the noise not fit KF by simulation to see whether algorithms’ switchover is practicable, if it does, the threshold is defined integrating the real situation; and puts it into the real video. The result shows that it is feasible of optimizing the tracking mode between KF and PF.

Key words: new tracking mode, Kalman Filter(KF), Particle Filter(PF), performance evaluation, algorithm switchover