This repository contains the Matlab code of a batch implementation of the trajectory Poisson Multi-Bernoulli mixture multi-object tracker using Markov chain Monte Carlo sampling.
The implementation is described in
Y. Xia, Á. F. García-Fernández, and L. Svensson, "Markov chain monte carlo multi-scan data association for sets of trajectories," in IEEE Transactions on Aerospace and Electronic Systems, 2024, available at https://arxiv.org/abs/2312.03423.
The trajectory Poisson multi-Bernoulli mixture filter is described in
K. Granström, L. Svensson, Y. Xia, J. Williams, Á. F. García-Fernández, "Poisson multi-Bernoulli mixture trackers: Continuity through random finite sets of trajectories," in 2018 21st International Conference on Information Fusion (FUSION), available at https://arxiv.org/abs/1812.05131 and https://arxiv.org/abs/1912.08718.
The tracking performance is evaluated using the trajectory generalised optimal subpattern-assignment (TGOSPA) metric, presented in
Á. F. García-Fernández, A. S. Rahmathullah, L. Svensson, "A metric on the space of finite sets of trajectories for evaluation of multi-target tracking algorithms," in IEEE Transactions on Signal Processing, 2020, available at https://arxiv.org/abs/1605.01177.
NOTE: implementations assign2DByCol.m and kBest2DAssign.m are copied from the GitHub repository https://github.com/USNavalResearchLaboratory/TrackerComponentLibrary.