| 余孟泽,刘正熙,骆键,杨丹.基于改进粒子滤波的鲁棒目标跟踪算法[J].光电子激光,2011,(5):766~771 |
| 基于改进粒子滤波的鲁棒目标跟踪算法 |
| Robust object tracking algorithm based on improved particle filter |
| |
| DOI： |
| 中文关键词: 目标跟踪 粒子滤波(PF) mean shift 变异重采样 |
| 英文关键词:object tracking particle filter mean shift mutate resample |
| 基金项目:国家“863”计划资助项目(2009AA01Z332) |
|
| 摘要点击次数: 401 |
| 全文下载次数: 25 |
| 中文摘要: |
| 为了克服样本贫化现象导致的滤波发散,本文对重采样后的粒子进行有方向性的变异操作,在增加样本集的多样性同时使粒子集更集中均匀的分布在目标的邻域。同时把Mean Shift算法引入粒子滤波(PF)框架中,对PF估计结果迭代得到最优的目标状态,并用迭代得到的状态值控制粒子变异的方向。仿真实验表明,本文提出的方法具有更高的估计精度,并且在目标状态突变时能够有效跟踪。 |
| 英文摘要: |
| In order to overcome particle impoverishment of the particle filter,this paper embeds the mutation operator into the resample step.After the mutation operator,the particles are more diversified and closer to the object′s real state.Then,mean shift algorithm is incorporated into the particle filter to get optimal object location and adjust the distribution of particles.The simulation results show that the proposed method can get more accurate estimation results and even when object motion is dras... |
| 查看全文 下载PDF阅读器 |
| 关闭 |
|
|
|
|
|