一种实用的融合方法在辅助粒子滤波中的应用  

Practical Data Fusion Algorithm Applied in Auxiliary Particle Filtering

作　　者：余熙[1] 张天骐[1] 白娟[1] 魏世朋[1] 

机构地区：[1]重庆邮电大学信号与信息处理重庆市重点实验室,重庆400065

出　　处：《计算机仿真》2012年第2期198-203,240,共7页Computer Simulation

基　　金：国家自然科学基金项目(61071196);国家自然科学基金-中物院NSAF联合基金项目(10776040);教育部新世纪优秀人才支持计划(NCET-10-0927);信号与信息处理重庆市市级重点实验室建设项目(CSTC;2009CA2003);重庆市自然科学基金项目(CSTC;2009BB2287;CSTC;2010BB2398;CSTC;2010BB2411)

摘　　要：研究机动目标跟踪优化问题,传统的扩展卡尔曼滤波(Extended Kalman Filtering,EKF),无迹卡尔曼滤波(UnscentedKalman Filtering,UKF)跟踪机动目标强非线性系统,滤波精度有限,并存在噪声。为了提高跟踪滤波的精准性和可靠性,提出建立一种协同转弯非线性模型,将辅助粒子滤波(Auxiliary Particle Filtering)和数据融合相结合的跟踪滤波算法。辅助粒子滤波在重采样前对各个粒子权值依据似然函数进行修正,同时采用一种将多个滤波器结合的实用的数据融合方法,可以进一步提高滤波精度。经仿真表明,算法能减小运算量,并能很好的提高跟踪滤波精度。The filtering for target tracking was discussed in the paper in order to find a method with better accuracy and reliability than extended kalman filtering and unscented kalman filtering. This paper built a non - linear system model, turn model. Then a method of auxiliary particle filters combined with data fusion was proposed. Auxiliary particle filtering for character of the turn model modified every particle before resampling according to likelihood func- tion, which can get the certain accuracy of filtering with fewest particles. In order to improve the filtering result further, the multi - auxiliary particle filters with a practical data fusion algorithm was used and it is efficient in turn model and can get good result through computer simulation.

关 键 词：似然函数 多辅助粒子滤波器 数据融合 转弯模型 目标跟踪 

分 类 号：TN953[电子电信—信号与信息处理]