航空学报 > 2016, Vol. 37 Issue (5): 1603-1613   doi: 10.7527/S1000-6893.2015.0209

一种基于最小二乘拟合的数据关联算法

王聪1,2, 王海鹏2, 熊伟2, 何友2   

  1. 1. 飞行器测控与通信教育部重点实验室, 重庆 400044;
    2. 海军航空工程学院 信息融合技术研究所, 烟台 264001
  • 收稿日期:2015-06-08 修回日期:2015-07-22 出版日期:2016-05-15 发布日期:2015-08-18
  • 通讯作者: 王聪,Tel.:0535-6635877 E-mail:congnavy@hotmail.com E-mail:congnavy@hotmail.com
  • 作者简介:王聪,男,博士研究生。主要研究方向:目标跟踪和航迹关联。Tel:0535-6635877 E-mail:congnavy@hotmail.com;王海鹏,男,博士,讲师。主要研究方向:群目标跟踪和航迹关联。Tel:0535-6635811 E-mail:armystudent@sohu.com;熊伟,男,博士,教授。主要研究方向:目标跟踪和误差配准。Tel:0535-6635807 E-mail:xiongweimail@tom.com;何友,男,中国工程院院士,教授。主要研究方向:雷达信号处理和信息融合。 E-mail:heyoumail@sohu.com
  • 基金资助:

    飞行器测控与通信教育部重点实验室开放基金(CTTC-FX201302)

Data association algorithm based on least square fitting

WANG Cong1,2, WANG Haipeng2, XIONG Wei2, HE You2   

  1. 1. Key Lab. for Spacecraft TT & C and Communication under the Ministry of Education, Chongqing 400044, China;
    2. Institute of Information Fusion, Naval Aeronautical and Astronautical University, Yantai 264001, China
  • Received:2015-06-08 Revised:2015-07-22 Online:2016-05-15 Published:2015-08-18
  • Supported by:

    Open Foundation for Key Lab. for Spacecraft TT & C and Communication under the Ministry of Education (CTTC-FX201302)

摘要:

针对点航关联在多目标跟踪中精度与实时性难兼顾的问题,提出了一种基于最小二乘拟合的点航关联算法。首先采用滑窗将历史航迹截断,采用最小二乘法在不同维度分别拟合、外推融合航迹历史信息条件下的航迹点,增加外推点的多样性及信息量。同时定义了5种全概率关联事件,提取传统滤波方法的预测点,将拟合外推点与滤波预测点融合,使归属判决更加准确。最后分别推导了不同事件发生时的状态更新方程与误差协方差更新方程,给出了其中参数的确定方法。经仿真数据验证,与经典的最近邻域法和联合概率数据互联算法相比,所提算法能够更好地兼顾精度与实时性,且计算复杂度较低,易于工程实现。

关键词: 最小二乘, 数据关联, 目标跟踪, 曲线拟合, 信息融合

Abstract:

Focusing on the hard problem of the balance between accuracy and real-time performance in multiple target tracking, a data association algorithm based on least square fitting method is proposed in this paper. Firstly, the tracking in sliding window is used to predict the next state by least square fitting respectively in different dimensions, which brings more history information to the attribution judgment. Then, cooperating with the prediction point of filter update, the next real position is judged by five defined probability events, which make the judgment of association more accurate. Finally, the state update equations and covariance are deduced in different events and the method to determine the parameters is given. The simulation results show that compared with the nearest neighbor algorithm and joint probabilistic data association algorithm, the proposed algorithm can be better in the balance of real-time and accuracy with low computational complexity, which is easy to implement in engineering practice.

Key words: least square, data association, target tracking, curve-fitting, information fusion

中图分类号: