基于动态网格密度的SNN聚类的ET-GM-PHD滤波算法

ET-GM-PHD Filtering Algorithm Based on Dynamic Grid Density SNN Clustering

针对不同扩展目标产生的量测密度差异较大时,多扩展目标高斯混合概率假设密度(ET-GM-PHD)量测集划分困难,计算量繁重的问题,提出了一种基于动态网格密度的SNN相似度的量测划分算法。首先利用动态网格技术对量测数据进行预处理,减小量测中的杂波干扰;而后采用共享最近邻(SNN)相似度对处理后的观测值进行量测划分。经过仿真结果分析,文中提出的算法相较于传统算法,减少了运行时间,提升了跟踪的稳定性。

In view of the large difference of measurement density produced by different extended targets,the problem of the multi extended target Gauss mixture probability hypothesis density( ET-GM-PHD) measurement set is difficult and the computational complexity is heavy,and a SNN similarity measurement division algorithm based on dynamic grid density is proposed. First,the dynamic grid technology is used to preprocess the measured data,and the clutter interference in the measurement is reduced,and then the shared nearest neighbor( SNN)similarity is used to measure the measured values. The simulation results show that the proposed algorithm reduces the running time and improves the tracking stability compared with the traditional algorithm.