Track-before-detect algorithm based on dynamic programming for multi-target tracking

A new dynamic programming based track before detect (DP-TBD) algorithm is presented to detect and track multi-target when SNR is low and target number is unknown. The technique combines the possible target state estimation using PDA with standard DP-TBD algorithm. And with the assumption that one measurement can originate from one target or clutter, the track separation phenomenon can be minimized and the target number can be obtained. Since the like-ratio function is not included in the merit function, the new method can be used without the priori information of the noise. Simulation results reveal that the new DP-TBD method has good performance when detecting target number and recovering target trajectories.