摘要
针对拓展目标概率假设密度滤波器采用的量测集合所有可能分割方式在实际中几乎不能够实现的问题,提出了一种采用有限混合模型的量测集合近似分割算法,对所有可能分割方式进行近似处理。算法利用有限混合模型拟合量测集合以实现对量测集合的分割,首先利用期望极大化算法极大似然估计混合参数,然后利用量测来源的条件概率分割量测集合,最后以二维场景为例进行了仿真实验。仿真结果表明:新算法在所有时刻上的最优子模式分配和混合分量数目均小于现有的典型量测集合分割算法;在拓展目标跟踪性能上,新算法具有更好的多拓展目标跟踪性能。
An approximate partition algorithm of measurement sets is proposed to overcome the problem that it is impossible to implement all the possible partitions of a measurement set in density filters with extended object probability hypothesis,and the algorithm bases on a finite mixture model.The finite mixture model is used to fit the measurement set and then the partition of the measurement set is implemented.The expectation maximization algorithm is employed to obtain the maximum likelihood estimation of mixture parameters.Then,the conditional probability of the measurement source is applied in partitioning the measurement set.Simulation results show that the proposed algorithm is superior to typical partition algorithm of measurement sets in extended object tracking.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2014年第9期19-23,29,共6页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(61304261)
江苏大学高级人才启动基金资助项目(12JDG076)
关键词
拓展目标跟踪
概率假设密度
量测集合分割
有限混合模型
extended object tracking
probability hypothesis density
measurement set partition
finite mixture model
