一种基于粒子滤波的联合跟踪与分类算法*

作者：申屠晗，郭云飞，薛安克 日期：2010-06-21/span> 浏览：3447 查看PDF文档

一种基于粒子滤波的联合跟踪与分类算法*
申屠晗，郭云飞，薛安克
（杭州电子科技大学 信息与控制研究所,浙江 杭州 310018）

摘要：针对纯运动学信息联合跟踪与分类问题，提出了一种基于混合无味粒子滤波的联合跟踪与分类算法。在传统粒子滤波联合跟踪与分类算法的基础上，通过采用无味变换，利用多个无味卡尔曼滤波器给出更高质量的粒子建议分布，提高整个算法的性能。理论分析和仿真结果都表明，与传统粒子滤波联合跟踪与分类算法相比，该算法无论在跟踪精度还是在分类正确率上都有明显的提高。
关键词：联合跟踪与分类；贝叶斯估计；混合无味粒子滤波；无味变换
中图分类号：TP391文献标识码：A文章编号：1001-4551(2010)03-0041-04

A joint tracking and classification algorithm based on particle filtering

SHEN Tuhan, GUO Yunfei, XUE Anke
(Institute of Information and Control, Hangzhou Dianzi University, Hangzhou 310018, China)

Abstract: In order to cope with the joint tracking and classification(JTC) problem, a new mixture unscented particle joint tracking and classification algorithm (MUPFJTC) was proposed. Based on traditional mixture unscented particle joint tracking and classification algorithm(MPFJTC), by adopting the methods of unscented transform(UT), several unscented Kalman filters(UKF) were designed in order to get higher quality particle distributions. Mathematical analysis and simulation results confirm that the MUPFJTC algorithm can achieve better estimation than common MPFJTC algorithm.
Key words: joint tracking and classification(JTC); Bayesian estimation; mixture unscented particle joint tracking and classification algorithm(MUPFJTC); unscented transform(UT)