基于改进KFCM聚类的数据关联算法  

A Data Association Algorithm Based on Improved KFCM Clustering

作　　者：施浩琴[1] 周德云[1] 张堃[1] 

机构地区：[1]西北工业大学,西安710129

出　　处：《电光与控制》2012年第4期13-17,58,共6页Electronics Optics & Control

基　　金：航空科学基金(20080553019)

摘　　要：针对传统的基于模糊C-均值(FCM)聚类的数据关联算法存在的缺陷,提出了一种基于改进核函数模糊C-均值(KFCM)聚类的数据关联算法。该算法以改进的KFCM聚类为基础,通过放宽KFCM聚类的约束条件来增强系统的鲁棒性,并引入信息熵自动确定目标数以作为数据关联的前期准备,再将改进的KFCM聚类算法引入JPDA算法,通过避免对联合事件的概率计算和对确认矩阵的拆分,以实现数据的正确关联和对多目标的实时跟踪。仿真结果表明算法有效可行。To overcome the shortcomings of traditional data association algorithm based on fuzzy C-mean （FCM） clustering,a novel data association algorithm based on improved kernel fuzzy C-mean （KFCM） clustering was proposed. The robustness of system was improved by loosing the constraints of clustering. Then entropy was integrated into KFCM clustering for determining the number of targets automatically, as the preparation of data association. After that the improved KFCM clustering algorithm was introduced in Joint Probabilistic Data Association （JPI）A）. By avoiding the probability calculation of composite events and matrix splitting, the proposed algorithm can implement correct data association and real-time tracking for multiple targets. Simulation results show that the proposed algorithm is rational and valid.

关 键 词：数据关联 目标跟踪 信息熵 核函数 模糊C-均值聚类 

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]