摘要
针对概率假设密度滤波器对新生目标的跟踪需要假设新生目标先验已知,其先验假设不合理,提出一种基于泊松多伯努利混合滤波器的新生目标跟踪方法。该方法根据泊松多伯努利混合滤波器共轭先验性质,获得当前量测为泊松过程和多伯努利混合过程的线性组成,并且将量测分别建立成新生目标量测和存活目标量测。以广义最优模式分配函数作为算法检测标准,实验结果表明,该方法对新生目标数目和目标位置的估计准确性均有较大提高。
Aimed at the problem of probability hypothesis density filter, which needs to assume that the new birth target is known a priori and its prior assumption is unreasonable, a new birth target tracking method based on Poisson multi-Bernoulli mixture filter is proposed. According to the conjugate prior property of Poisson multi-Bernoulli mixture filter, the method obtained the linear composition of the current measurement as Poisson process and multi-Bernoulli mixture process, and established the measurement as new birth target measurement and survival target measurement respectively. The generalized optimal subpattern assignment function was used as the algorithm detection standard. Experimental results show that the method greatly improves the estimation accuracy of the number of new birth targets and the target position.
作者
鑑美玉
柳晓鸣
Jian Meiyu;Liu Xiaoming(College of Information Science and Technology,Dalian Maritime University,Dalian 116026,Liaoning,China)
出处
《计算机应用与软件》
北大核心
2022年第12期292-297,共6页
Computer Applications and Software
基金
福建海事局基金项目(2018Z0093)。
关键词
多目标跟踪
随机有限集
泊松多伯努利混合
新生目标
Multi-target tracking
Random finite sets
Poisson multi-Bernoulli mixture
Birth target
