Computer Engineering and Applications ›› 2019, Vol. 55 ›› Issue (5): 8-17.DOI: 10.3778/j.issn.1002-8331.1809-0242

Previous Articles     Next Articles

Survey of Particle Filter Target Tracking Algorithms

ZAN Meng’en, ZHOU Hang, HAN Dan, YANG Gang, XU Guoliang   

  1. School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing 100044, China
  • Online:2019-03-01 Published:2019-03-06

粒子滤波目标跟踪算法综述

昝孟恩,周  航,韩  丹,杨  刚,许国梁   

  1. 北京交通大学 电子信息工程学院,北京 100044

Abstract:

With the development of artificial intelligence science, target tracking has become a hotspot for domestic and foreign scholars. In recent years, many target tracking algorithms have been proposed. Among them, the classical Kalman filtering algorithm is often used in the target tracking field. However, in the actual situation, the target tracking process often involves nonlinear non-Gaussian problems. As the particle filtering algorithm has better performance in non-Gaussian nonlinear systems, it is introduced into the field of target tracking research. In view of the problems of poor tracking accuracy and low real-time performance of particle filtering algorithm, many domestic and foreign scholars have proposed many improved methods. In this paper, the basic ideas of related improved methods are introduced from three aspects:feature fusion, algorithm fusion and adaptive particle filtering. The development direction of particle filtering algorithm in target tracking field is prospected.

Key words: target tracking, particle filtering algorithm, resampling, importance sampling, feature fusion, adaptive particle filter

摘要: 随着人工智能科学的发展,目标跟踪成为中外学者研究的热点,近年来很多目标跟踪算法相继被提出,其中,经典的卡尔曼滤波算法常被用于目标跟踪领域。然而,在实际情况中,目标跟踪过程常涉及到非线性非高斯问题,由于粒子滤波算法在非线性非高斯系统中有较好的性能,因此将其引入目标跟踪研究领域。针对粒子滤波算法存在的跟踪精度差、实时性不高等问题,近年来国内外学者提出很多改进方法。从特征融合、算法融合和自适应粒子滤波三个方面介绍了相关改进方法的基本思想,展望了粒子滤波算法在目标跟踪领域的发展方向。

关键词: 目标跟踪, 粒子滤波, 重采样, 重要性采样, 特征融合, 自适应粒子滤波