| 许彦伟,侯朝焕,李军,郝程鹏.恒虚警率采样粒子滤波技术及其应用研究[J].,2013,32(4):320-324 |
| 恒虚警率采样粒子滤波技术及其应用研究 |
| Research on particle filter based on constant false alarm rate sampling |
| 投稿时间：2013-06-27 修订日期：2013-07-01 |
| 中文摘要: |
| 粒子滤波是一种基于蒙特卡洛思想的非线性、非高斯滤波器,其一般采用重要性采样进行粒子采样。但重要性采样容易出现粒子退化现象。解决粒子样本退化问题一般采用重采样。重采样虽然解决了样本的退化问题,同时又引入了采样贫瘠问题。本文根据海洋混响的统计特性和混响中目标的恒虚警率检测原理,提出了恒虚警率采样粒子滤波技术,恒虚警率采样粒子滤波技术使采样粒子尽可能集中在目标附近,有效地描述目标后验概率,降低了粒子数,减小了计算量。本文将此技术应用到海洋混响中的声纳目标跟踪中,既解决了传统卡尔曼滤波在声纳目标跟踪中的非线性、非高斯问题,又解决了粒子滤波的粒子退化及采样贫瘠问题。文中对高分辨率声纳目标数据进行了滤波跟踪,结果验证了本文方法的有效性。 |
| 英文摘要: |
| Particle Filter is a non-linear and non-Gaussian filter based on Monte Carlo Methods. Important sampling is usually adopted to sample the particles, but it is difficult to tackle the problem of particle degeneracy. Re-sampling is proposed to tackle the degeneracy phenomenon, however at the meantime the sample impoverishment that leads to the loss of effectiveness and diversity among the samples arises. In this paper, an improved particle filter based on constant false alarm rate (CFAR) sampling is proposed. The effective sampling method clusters the particles around the target as much as possible, thus better representing the posterior density, improving the sampling efficiency and decreasing the computational load. The proposed Particle Filter is used in sonar target tracking, which overcomes the shortcomings of traditional Kalman Filter and classical particle filter. Experimental results based on a set of high resolution sonar data demonstrate the effectiveness of the proposed particle filter. |
| DOI：10.11684/j.issn.1000-310X.2013.04.011 |
| 中文关键词: 状态估计,重采样,恒虚警率采样,粒子滤波 |
| 英文关键词: State estimation, Re-sampling, Constant false alarm rate sampling, Particle filter |
| 基金项目:国家自然科学基金(61172166)资助项目、中国博士后科学基金(2013M530744)资助项目 |
|
| 摘要点击次数: 2351 |
| 全文下载次数: 2039 |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
