Multitarget Tracking Before Detection via Probability Hypothesis Density Filter

Tracking-before-detection (TBD) is well suitable for radar detection and target tracking of low-observable objects because it makes full use of the raw sensor data without using the threshold and takes advantage of the gains of time integration. The main difficulty in the TBD is that the measurement is a highly nonlinear function of the target state and the associations of measurements are hard to be built when multiple targets are presented. Probability hypothesis density (PHD) filter is regarded as an efficient solution to multitarget tracking problems. To deal with multitarget TBD problem, a classical PHD filter, with "standard" multitarget measurement model, is proposed in this paper. The advantage of proposed filter is analyzed compared with traditional multitarget particle filter. Numerical simulations show our approach has better performance in estimated accuracy and precision of position, velocity and number of targets than multitarget particle filter.