计算机工程与应用 ›› 2014, Vol. 50 ›› Issue (17): 205-209.

• 信号处理 • 上一篇    下一篇

高斯粒子群优化粒子滤波检测前跟踪算法

汲清波,耿丽群,任  超   

  1. 哈尔滨工程大学 信息与通信工程学院,哈尔滨 150001
  • 出版日期:2014-09-01 发布日期:2014-09-12

Track before detect algorithm based on Gaussian particle swarm optimization particle filter

JI Qingbo, GENG Liqun, REN Chao   

  1. College of Information and Communication Engineering, Harbin Engineering University, Harbin 150001, China
  • Online:2014-09-01 Published:2014-09-12

摘要: 针对低信噪比时标准粒子滤波对弱小目标的检测与跟踪时存在的粒子贫乏、跟踪精度对粒子数目要求高等问题,提出一种基于高斯粒子群优化粒子滤波的弱小目标检测前跟踪算法。利用高斯粒子群优化算法优化重采样后的粒子集,使粒子集朝着后验概率密度分布取值较大的区域运动,增加粒子的多样性,克服了粒子贫乏问题,并在保证跟踪精度的前提下降低了跟踪所需要的粒子数目,提高了标准粒子滤波算法的检测和跟踪性能。同时,建立了检测前跟踪系统的观测模型和系统模型,对基于标准粒子滤波检测前跟踪算法和优化算法进行仿真,仿真实验结果表明高斯粒子群优化粒子滤波的检测前跟踪算法相比基于标准粒子滤波的检测前跟踪算法具有更好的检测与跟踪性能。

关键词: 弱小目标, 检测前跟踪, 高斯粒子群优化算法, 粒子滤波算法

Abstract: Considering the particle impoverishment and the large particle sample size for the high tracking accuracy in dim target track before detect algorithm based on standard particle filter (PFTBD), a dim target track before detect algorithm based on Gaussian particle swarm optimization particle filter (PSOPFTBD) is proposed. Gaussian particle swarm optimization algorithm is applied to optimize the particles after resampling, which makes the particles move towards the larger values of posterior density function. The PSOPFTBD algorithm increases the diversity of the particles, overcomes the particle impoverishment, and the particle sample size for accurate state estimation is also reduced. As a result, the detection and tracking performance is improved. The PSOPFTBD algorithm is tested on a tracking and detection mathematical model, and the results are compared with the results of the standard PFTBD algorithm. The simulation results show that the PSOPFTBD algorithm outperforms the PFTBD algorithm.

Key words: dim target, track before detect, Gaussian particle swarm optimization, particle filter