带反馈多传感器模糊最大熵单目标跟踪算法  

Fuzzy tracking algorithm with feedback based on maximum entropy principle

作　　者：刘智[1] 陈丰[1] 黄继平[1] 

机构地区：[1]重庆理工大学计算机科学与工程学院,重庆400054

出　　处：《计算机应用研究》2012年第7期2467-2469,2508,共4页Application Research of Computers

基　　金：重庆市教委科学技术研究资助项目(KJ110805);重庆理工大学青年基金计划资助项目(2011ZQ9)

摘　　要：针对矩阵加权融合算法计算量大、传感器数量不易扩充的特点,提出了一种带反馈的模糊最大熵融合算法。该算法采用模糊C-均值算法和最大熵原理计算状态向量中每一分量的权值,不但从整体考虑各分量对融合估计的影响,而且减少了复杂的矩阵运算过程,实时性较好。与矩阵加权算法相比,该融合算法还具有容易扩充的特点,能够直接应用于传感器数量大于2时的融合计算。实验仿真结果表明,融合估计的准确性与矩阵加权融合算法基本一致,算法的有效性得到了验证。Aiming at the disadvantages of high computation overhead and bad extensibility in matrix weighted fusion methods,this paper proposed a multisensor fusion algorithm with feedback based on fuzzy C-means（FCM） clustering and maximun entropy principle（MEP）.This algorithm combined FCM and MEP to calculate fusion matrix weight of state vector considering every components integratedly.What’s more,this algoritm had a good real-time performance due to less matrix computation and good extensibility which showed it could directly be applied into tracking system comprising more than two sensors.Experiments and results reveal that the tracking accuracy of fusion estimate is consistent with that of matrix weighted fusion methods.

关 键 词：信息融合 多传感器 模糊C-均值 最大熵原理 加权矩阵 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论] TP301.6[自动化与计算机技术—计算机科学与技术]