基于关联的自适应新生目标强度CPHD滤波

量测驱动的自适应新生目标强度基数概率假设密度(adaptive target birth intensity cardinalized probability hypothesis density,ATBI-CPHD)滤波器可以在新生目标强度未知的情况下进行多目标跟踪,然而该方法利用所有量测产生新生目标,没有考虑关联问题。为此,本文提出了一种基于数据关联的改进算法。首先,给出了ATBI-CPHD在高斯混合CPHD(Gaussian mixture CPHD, GMCPHD)框架下的实现。其次,在GMCPHD滤波框架下采用一种基于量测标签的方法进行量测估计关联,并引入高斯元标签进行航迹保持,在此基础上提出了一种航迹管理方法。最后采用量测波门进行量测量测关联,利用关联后的量测产生新生目标。仿真结果表明，该算法可以在提高跟踪效果的同时提升计算效率。

Association based adaptive target birth intensity CPHD filter

The measurement-driven adaptive target birth intensity cardinalized probability hypothesis density (ATBI-CPHD) filter can track multiple targets without prior information of target birth intensity. The association between measurements and estimates is almost not considered in the intensity update stage, however, it is obviously inefficient. An efficiency-enhanced ATBI-CPHD filter based on the association is presented to correct it. Firstly, the ATBI-CPHD filter is implied in the Gaussian mixture framework. Secondly, a novel measurement-estimate association method is presented, and the search scheme using the measurement gate is designed to find out the most likely newborn measurements from those disassociated measurements and create the newborn targets. And a track management method based on the Gaussian term label and the measurement label is proposed and applied to the ATBI-CPHD filter. The simulation results prove that, compared with the traditional methods, the proposed approach leads to better efficiency and estimate accuracy.