一种基于改进FCM聚类联合概率数据关联算法  

A Joint Probabilistic Data Association Algorithm Based on Improved FCM Clustering

作　　者：孙炜[1] 吕辉[1] 白剑林[1] 

机构地区：[1]空军工程大学导弹学院,陕西三原713800

出　　处：《电光与控制》2010年第8期11-14,共4页Electronics Optics & Control

基　　金：国家自然科学基金(60601016)

摘　　要：针对多目标跟踪系统中的数据关联问题,提出一种基于改进FCM聚类联合概率数据关联算法(FJPDA)。该算法将改进的FCM聚类方法引入JPDA算法中,避免了对联合事件的概率计算,也避免了对确认矩阵拆分造成的计算量组合爆炸现象,实现了量测与航迹的关联,继而实现对多目标的实时跟踪。仿真结果表明算法简单有效,与JPDA算法相比,在跟踪性能相当的前提下,算法的复杂度和实时性得到了明显的改善。For data association of multi-target tracking system,an improved FCM clustering based algorithm for Joint Probabilistic Data Association（JPDA） is proposed.The algorithm introduces FCM clustering in JPDA,thus can not only avoid the probability calculation of composite event,but also avoid the combinatorial explosion of computation in matrix splitting.The association between measurement and track is realized,and the real-time tracking of multi-target is achieved accordingly.Simulation results indicate the validity of the algorithm.In comparison with JPDA,the complexity is reduced and the real-time performance is improved evidently for the equivalent tracking performance.

关 键 词：目标跟踪 数据关联 隶属度 模糊聚类 

分 类 号：V271.4[航空宇航科学与技术—飞行器设计] N941.5[自然科学总论—系统科学]