Multi-sensor Multi-Target Tracking Using Labelled Random Finite Sets with Homography Data

This paper proposes a solution for multi-sensor multi-Target tracking with homography data using the labelled random finite set with a top-down Bayesian recursion formulation. The proposed method encapsulates multi-Target state motion, appearance and disappearance and all aspects of noise, detection and association uncertainty from multiple sensors. This technique naturally incorporates the fusion of multi-sensor measurements to improve the fidelity of multi-Target trajectories estimation. A linear Gaussian multi-Target model with simulated homography data from multiple sensors is undertaken for verification.