H-PMHT Track-Before-Detect Method Using Markov Chain Monte Carlo Particle Filter
Jinping Sun, Ying Lu, Zhiguo Zhang · 2018
This paper addresses the problem of detecting and tracking dim targets in low signal-to-noise ratio (SNR) environment. The histogram probability multi-hypothesis tracking (H-PMHT) algorithm based on particle filter, as an effective track-before-detect (TBD) method, is proposed to solve this task. Since the classical resampling particle filter loses the diversity of particles after resampling and needs heavy computation load and large storage capacity in the multi-target scenario, we apply Markov chain Monte Carlo (MCMC) particle filter to H-PMHT method. The efficiency of the algorithm is guaranteed by increasing the diversity of particles. Finally, the detection performance of the H-PMHT algorithm based on MCMC particle filter and its superiority on computation complexity are verified through simulation experiments.