基于STGP-ETCBMeMBer滤波器的扩展目标跟踪算法  

Non-rigid Extended Target Tracking Based on STGP-ETCBMeMBer Filter

作　　者：吴孙勇[1,2] 吕晓燕 薛秋条 周于松[1] WU Sunyong;LYU Xiaoyan;XUE Qiutiao;ZHOU Yusong(School of Mathematics and Computing Science,Guilin University of Electronic Technology,Guilin,Guangxi 541004,China;Guilin Key Laboratory of Cryptography and Information Security,Guilin,Guangxi 541004,China)

机构地区：[1]桂林电子科技大学数学与计算科学学院,广西桂林541004 [2]广西密码学与信息安全重点实验室,广西桂林541004

出　　处：《信号处理》2023年第7期1319-1330,共12页Journal of Signal Processing

基　　金：国家自然科学基金(62263007,62061010,62161007);中央引导地方科技发展基金项目(2022YZX2001);广西科技厅项目(AA19182007);广西密码学与信息安全重点实验室研究课题(GCIS202132);桂林电子科技大学数学与计算科学学院研究生创新项目(2022YJSCX02)。

摘　　要：外形估计是扩展目标跟踪的难点之一,不精确的建模方法会导致较差的估计效果。针对不规则多扩展目标的跟踪问题,本文提出了一种基于时空高斯过程模型的势平衡多扩展目标多伯努利(Spatio-Temporal Gaussian Process Cardinality Balanced Multi-Extended Target Multi-Target Multi-Bernoulli, STGP-ETCBMeMBer)滤波算法。首先,利用随机有限集方法将增广后的多扩展目标状态集合和量测集合分别建模为多伯努利随机有限集合和泊松随机有限集合,并在此基础上利用STGP方法对星凸形扩展目标的量测源建模,提高算法对扩展目标的外形估计精度。之后,在算法更新阶段,假设同一目标量测子集对应的多个似然函数服从高斯分布,推导得到了STGPETCBMeMBer滤波算法的高斯混合(Gaussian Mixture, GM)实现。最后,通过构造仿真对比实验验证了所提算法的有效性,仿真结果表明所提算法在扩展目标的外形估计上有更精确的效果。Shape estimation is one of the difficult aspects of extended target tracking,and inaccurate modeling methods can lead to poor estimation results.For the tracking problem of irregular multi-extended targets,a Cardinality Balanced Multi-Extended Target Multi-Bernoulli(STGP-ETCBMeMBer)filtering algorithm is proposed based on the Spatio-Temporal Gaussian Process(STGP)model.First,the augmented multi-extended target state set and the measurement set are modeled as multi-Bernoulli random finite sets and Poisson random finite sets,respectively,using the random finite set method,and on this basis,the STGP method is used to model the measurement sources of the star-convex extended target to improve the algorithm’s shape estimation accuracy of the extended target.After that,in the algorithm update stage,multiple likelihood functions corresponding to the same target measurement subset are assumed to obey Gaussian distribu⁃tion,and closed form solution realized by Gaussian mixture(GM)implementation is derived.Finally,the effectiveness of the proposed algorithm is verified by constructing simulation comparison experiments,and the simulation results show that the proposed algorithm has a more accurate effect on the shape estimation of extended targets.

关 键 词：扩展目标 时空高斯过程 CBMeMBer滤波器 高斯混合 

分 类 号：TN953[电子电信—信号与信息处理]