UKF滤波器的强跟踪性研究  

Research on Strong Tracking Features of Unscented Kalman Flter

作　　者：冯志全[1] 孟祥旭[1] 蔺永政[2] 

机构地区：[1]山东大学计算机科学与技术学院,山东济南250100 [2]济南大学信息科学与工程学院,山东济南250022

出　　处：《小型微型计算机系统》2006年第11期2142-2145,共4页Journal of Chinese Computer Systems

基　　金：国家高技术研究发展计划项目(2003AA414310)资助;国家自然科学基金项目(60473103;60473127)资助.

摘　　要：与EKF滤波器相比较,UKF滤波器和强跟踪滤波器各自具有不同的特点,在基于计算机视觉的运动人手跟踪系统中往往需要将这些不同的特点相结合.本文揭示了UKF滤波器与强跟踪滤波器之间的关系.研究表明,在线性系统中,UKF滤波器是强跟踪滤波器;在非线性系统中,在一定的条件下,UKF是强跟踪滤波器,本文给出了一个充分条件.最后给出了部分实验结果.Strong Tracking Filter（STF） and Unscented Kalman Filter（UKF） has many unique and appealing features respectively, and how to combine these features into the moving hand tracking base on 2-D monocular videos is an open area. This paper is dedicated to explore the relationship between STF and UKF. The features of UKF are studied in the linear system and nonlinear system respectively and it is exposed that, in linear systems UKF is STF and in non-linear systems UKF is STF only if some sufficient conditions are satisfied which are also presented in this paper. All conclusions are proved in detail and the experimental results with real data demonstrate our conclusions.

关 键 词：卡尔曼滤波 无味卡尔曼滤波 人手跟踪 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]