基于粒子滤波的检测前跟踪技术(TBD-PF)是低信噪比环境下目标检测与跟踪的有效方法。针对传统的TBD-PF算法经重采样后容易导致粒子耗尽,从而跟丢目标的缺点。本文提出采用对权重较低的部分"存活"粒子用"新生"粒子将其替换,且对重采样后的粒子实施马尔科夫链蒙特卡洛(MCMC)移动步骤的粒子更新策略,在增加粒子多样性的同时保证了粒子的有效性。仿真实验分析了粒子数大小,预设门限的选取对检测性能的影响。实验结果表明,所提算法的检测与跟踪性能要优于传统的粒子滤波算法。
Particle filter-based for track-before-detect ( TBD) is an efficient approach for weak target detection and track under low SNR environment. The resampling method is apt to induce collapse of particles and thus fails to detect target. An updating strategy that replaces the existing particles with low weights by new particles is proposed and a Markov chain Monte Carlo ( MCMC) move step is performed after resampling particles. This strategy can improve the diversity among the particles,and ensure that the particles are effective. Simulation analysis is given for the effect of the number of particles and selection of the detection threshold on target detection. Simulation results show that the proposed algorithm provides an improved performance for detecting and tracking weak targets compared with the conventional particle filter.