Multiple target tracking using track before detect algorithm

In this paper, we adapted the recursive TBD algorithm to track multiple targets in low SNR. This algorithm is important, for example, in closely spaced targets moving along the same direction. The different modes in the multiple target case correspond to the different number of target combination that may be present in the scene at each time. Thus, the multiple-mode multiple targets track before detect (MM-MT-TBD) algorithm is based on estimating the posterior probability density of the target states under different modes in order to consider all possible target existence combination at each time step. We implement the multi-mode multi-target track before detect (MM MT-PF-TBD) recursively using particle filtering to also incorporate nonlinear tracking models for moving targets. The performance of this algorithm is demonstrated using a simulated two targets as an example for image measurements.