【作者】 徐金豹；

【导师】 王从庆；

【作者基本信息】 南京航空航天大学 ， 模式识别与智能系统， 2008， 硕士

【摘要】 机动目标跟踪在国防科研、雷达信号处理及其他相关领域中是一个非常重要的研究课题,最近几十年来,国内外众多专家学者对之进行了深入的研究,并取得了丰硕的成果,其中部分研究成果已经成功应用于空中侦察与预警、弹道导弹防御、战场监视等军事领域,以及空中交通管制、交通导航、机器人视觉系统等民用领域。本论文主要对机动目标跟踪中的非线性滤波算法进行了系统深入的研究。首先系统地概述了机动目标跟踪的基本原理,讨论了几种常用的目标机动模型、基本滤波预测方法和数据关联算法。分析比较了多种非线性滤波算法和新的粒子滤波算法,实验结果表明代价参考粒子滤波算法有较好的滤波性能。其次为解决非线性非高斯条件下的机动目标跟踪问题,论文将代价参考粒子滤波算法与当前统计模型的优点相结合,提出了基于代价参考粒子滤波的“当前”统计模型自适应跟踪算法,仿真结果表明该算法用于解决机动目标跟踪问题是有效可行的;研究了基于代价参考粒子滤波的多模型跟踪算法,对该算法重采样过程随机性太大的缺陷做了改进,并通过仿真实验证明了改进算法的有效性和可行性;分析比较了多模型跟踪算法中模型数目选取对跟踪性能的影响,并对基于代价参考粒子滤波的两种跟踪算法的跟踪性能做了对比分析。最后针对多目标跟踪中的数据关联问题,阐述了几种典型的多目标数据关联算法,研究了另一种新的粒子滤波算法,即Rao-Blackwellised粒子滤波算法,然后将该算法应用于单机动目标的跟踪问题和多机动目标的数据关联问题,仿真结果表明,该算法有较高的跟踪精度和较好的实时性。论文还提出今后机动目标跟踪的发展方向。更多还原

【Abstract】 The maneuvering target tracking is an extremely important research topic in national defense, the processing of radar signals and other related areas. In recent years, target tracking technology has been widely studied, and has made plentiful and substantial achievements. Some research results have been widely applied to military fields, such as air reconnaissance and early warning, ballistic missile defense, battlefield surveillance, etc. Some research results have been applied to civil fields, such as air traffic control, traffic navigation and robot vision system, etc. This paper researches nonlinear filtering algorithms of maneuvering target tracking more systemic.Firstly, basic principle of maneuvering target tracking is summarized and some familiar target maneuvering models, basic filtering and data association algorithms are also discussed. Following the discussion, this paper analyzes and compares several nonlinear filtering algorithms, the simulation results demonstrate the Cost Reference particle filter has an excellence filtering performances.Secondly, in order to solve nonlinear and non-Gaussian maneuvering target tracking problems, this paper integrates the advantages of the Cost Reference particle filter with the Current Statistical Model, and proposes a new current statistical model adaptive tracking algorithm, and the simulation results demonstrate its availability for maneuvering target tracking. This paper also discusses the multiple models adaptive tracking algorithm based on Cost Reference particle filter, and analyzes the blemish that the over randomicity of resample algorithm, as a result this paper puts forward a improved algorithm and the simulation validates the availability and applicability of the improved algorithm. The effects on algorithm performances by the amount of selected models in the multiple models tracking algorithm are investigated, and the tracking performances of the two tracking algorithms based on Cost Reference particle filter are compared through separate simulation in this paper.Finally, aiming at the data association problems of multiple targets, this paper expatiates some representative algorithms of multiple targets data association. Another new Rao-Blackwellised Particle Filter algorithm is discussed in details, and then is applied to solve the single maneuvering target tracking problem and the data association problem of multiple targets. The simulation results demonstrate the algorithm has highly real-time performance and the tracking accuracy rate. This paper also presents the future development of maneuvering target tracking.更多还原