基于标签随机有限集滤波器的多扩展目标跟踪算法  

Multiple extended target tracking algorithm based on labeled random finite set filter

作　　者：曹倬[1] 冯新喜[1] 蒲磊[1] 王雪[1] 张琳琳 CAO Zhuo;FENG Xinxi;PU Lei;WANG Rue;ZHANG Linlin(Information and Navigation Institute, Air Force Engineering University, Xi＇an 710077, China;The Ministry of Basic Computer Application Teaching and Research Office, Airforce Dalian Communications Noncommissioned Officers School, Dalian 116600, China)

机构地区：[1]空军工程大学信息与导航学院,陕西西安710077 [2]空军大连通信士官学校基础部计算机应用教研室,辽宁大连116600

出　　处：《系统工程与电子技术》2018年第3期526-532,共7页Systems Engineering and Electronics

基　　金：国家自然科学基金(61571458)资助课题

摘　　要：针对现有随机有限集(random finite set,RFS)扩展目标滤波器不能输出航迹的问题,提出了基于标签RFS滤波器的多扩展目标跟踪算法。该算法首先采用随机超曲面模型将目标建模为星-凸扩展形态,然后利用标签策略表征集合中的离散元素,结合基于延迟逻辑的多假设跟踪理论,采用N次回扫策略对多帧量测进行平滑处理。仿真实验结果表明,该算法可以在目标跟踪过程中形成完整航迹并对目标扩展形态进行有效估计,特别是在低信噪比探测场景中,所提算法跟踪精度明显优于传统RFS滤波算法,进一步提高了滤波器的稳定性和有效性。A multiple extended target tracking algorithm based on labeled random finite set（RFS）filter is proposed to solve the problem that the existing algorithm based on the RFSs cannot output the target kinematic track.The proposed algorithm adopts the random hypersurface to model the extended target as star-convex extension,and then puts the elements in the sets labeled.Combining with the delay logic theory of multiple hypothesis tracking,we adopt the backtracking strategy to smooth multiframe measurements.Experimental results show that the proposed algorithm is able to output complete track and effectively estimate the kinematic state and target extension when tracking the target,especially in the low signal-to-noise ratio scenario,the tracking accuracy of the proposed algorithm is better than that of the traditional extended target filter based on the RFS theory,so that the stability and effectiveness of the filter can be improved.

关 键 词：目标跟踪 多扩展目标 标签随机有限集 航迹维持 

分 类 号：TN953[电子电信—信号与信息处理]