电子学报 ›› 2018, Vol. 46 ›› Issue (10): 2472-2479.DOI: 10.3969/j.issn.0372-2112.2018.10.022

• 学术论文 • 上一篇    下一篇

基于分治-贪心算法的高斯混合多观测站CPHD滤波器

曲长文1, 冯奇2, 李廷军1   

  1. 1. 海军航空工程学院电子信息工程系, 山东烟台 264001;
    2. 武警警官学院分队指挥系, 四川成都 610200
  • 收稿日期:2017-10-09 修回日期:2017-12-14 出版日期:2018-10-25
    • 作者简介:
    • 曲长文,男.1963年11月出生,山东济南人.教授、博士生导师.1983年、1995年在海军航空工程学院获得工学学士和工学硕士学位,2004年在海军工程大学获得工学博士学位.主要从事无源定位跟踪技术、信号与信息处理技术方面的研究;冯奇,男.1990年1月出生,河南焦作人.2011年、2013年在海军航空工程学院获得学士学位和硕士学位.现为博士研究生,主要研究方向为无源定位跟踪技术.E-mail:fengqi1109@163.com
    • 基金资助:
    • 国家自然科学基金 (No.91538201)

Multisensor CPHD Filter with Gaussian Mixture Implementation Based on Divide-Conquer and Greedy Algorithm

QU Chang-wen1, FENG Qi2, LI Ting-jun1   

  1. 1.Department of Electronic and Information Engineering, Naval Aeronautical and Astronautical University, Yantai, Shandong 264001, China;
    2.Department of Detachment Command, Officers College of PAP, Chengdu, Sichuan 610200, China
  • Received:2017-10-09 Revised:2017-12-14 Online:2018-10-25 Published:2018-10-25
    • Supported by:
    • National Natural Science Foundation of China (No.91538201)

摘要: 针对现有的多观测站概率假设密度滤波器实现中存在依赖观测站处理顺序、计算复杂度高等问题,文中提出一种基于分治-贪心算法的高斯混合多观测站势概率假设密度滤波器.假设观测站个数为s,每个观测站的量测个数为n,相对于暴力分析法,分治算法使得子集选取问题的计算复杂度从Ons)降到了Ons).此外,在线性高斯模型假设条件下,给出多观测站势概率假设密度滤波实现的具体步骤.仿真结果证明,本文实现方法不受观测站处理顺序的影响,分治-贪心近似实现方法与暴力分析法的跟踪性能相当,但是运算耗时大大降低,提高了算法实现及应用的可行性.

关键词: 多目标跟踪, 势概率假设密度, 分治算法, 贪心算法

Abstract: For the problem that the existing multisensor Probability Hypothesis Density (PHD) filters have limitations such as sensor order dependence or high computational requirements, a multisensor cardinalized probability hypothesis density (CPHD) filter with Gaussian mixture implementation based on Divide-conquer and Greedy algorithm is proposed in this paper. Especially the divide-conquer algorithm make the computation complex fall to O(ns) from O(ns), compared with the brute analysis algorithm, where s is the number of sensors and n is the number of observations of each sensor. In addition, the details of the implement of the multi-sensor CPHD filter are expressed in this paper. Finally, simulation results show that the algorithm proposed is not affected by the processing order, the proposed divide-conquer and greedy two-stage algorithm is equivalent to the tracking performance of the brute analysis, but the operation time greatly reduced, all of this improve the feasibility of implementation and application of multi-sensor CPHD filter.

Key words: multi-target tracking, cardinalized probability hypothesis density, divide-conquer algorithm, greedy algorithm

中图分类号: