Fast Generalized Labeled Multi-Bernoulli Tracking Algorithm Based on Box Particle Filtering

Abstract

For the case that the prediction and update steps of sequence Monte Carlo Generalized Labeled Multi-Bernoulli filter (SMC-GLMB) require pruning respectively which causes large amount of calculation and low operation efficiency, a fast GLMB algorithm based on box particle filter for target tracking is proposed. First, a new recursive equation is derived based on the combination of prediction and update step, then the box particle filter is used to approximate the probability density of single target state, finally we use the new recursive equation to update the probability density of target state. Simulation results show that our proposed algorithm can effectively estimate the state of target, and the computational efficiency is significantly improved compared with the traditional SMC-GLMB filter.