A Recursive Bayesian Method for Multi-Target Detection and Tracking Using Particle Swarms

Abstract

A recursive Bayesian method of multi-target detection and tracking (MTDT) is proposed. Two kinds of particle swarms, birth and tracking particle swarms, are employed to implement a recursive Bayesian filter for MTDT by two steps of update and resampling, where the resampling of the particle swarms is done by estimating their associated probabilities of target existence with a proposed method. Each particle swarm is designed to deal with only one target in the way of single target detection and tracking. The results of 100 times of Monte Carlo experiments show that it can effectively detect and track multiple targets with low SNR.