| 引用本文: | 鉴福升,徐跃民,阴泽杰.改进的多模型粒子滤波机动目标跟踪算法[J].控制理论与应用,2010,27(8):1012~1016.[点击复制] |
| JIAN Fu-sheng,XU Yue-min,YIN Ze-jie.Enhanced multiple model particle filter for maneuvering target tracking[J].Control Theory and Technology,2010,27(8):1012~1016.[点击复制] |
|
| 改进的多模型粒子滤波机动目标跟踪算法 |
| Enhanced multiple model particle filter for maneuvering target tracking |
| 摘要点击 2493 全文点击 1346 投稿时间：2009-06-16 修订日期：2009-09-20 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/j.issn.1000-8152.2010.8.CCTA090771 |
| 2010,27(8):1012-1016 |
| 中文关键词 粒子滤波 多模型 跟踪算法 粒子数 似然函数 |
| 英文关键词 particle filter multiple model tracking algorithm number of particles likelihood function |
| 基金项目 |
|
| 中文摘要 |
| 传统的多模型粒子滤波算法(MMPF)在机动目标跟踪中存在模型粒子数难以控制的问题. 本文提出了一种改进的多模型粒子滤波算法, 将模型估计和状态估计分开计算, 并用模型似然函数计算模型后验概率. 各模型的粒子数根据模型特性预先选定, 在递推过程中保持不变, 且模型间的粒子不进行交互, 减少了计算量. 仿真表明,同MMPF相比, 该算法能用较少的粒子数获得更好的跟踪精度. |
| 英文摘要 |
| Classical multiple model particle filters(MMPF) can not effectively control particles of each mode in maneuvering target tracking. An enhanced multiple model particle filter(EMMPF) algorithm is proposed, which estimates the mode and state independently. The posterior probability of each mode is updated with the associate likelihood function. The number of particles of each mode is given in advance according to the mode property and is kept constant in the recursion. Simulation shows that the EMMPF achieves better tracking accuracy with fewer particles. |