Abstract:

To solve the multi-target joint tracking and classification under clutter and miss detection, the model-class-matched probability hypothesis density (PHD) filter is proposed and combined with the transferable belief model (TBM) framework of multiple sensors, and it is recursively implemented by multiple kinematic radars and particle filtering. In our algorithm, the priori information of aircraft is estimated to replace the attribute sensor. At the prediction stage, the particles of the PHD filter are matched by model and class; the particle weights are updated after the sensors receive the measurement information; then, the targets’ posterior statemodelclass distribution is got by the corresponding weights of the particles; the targets’ state and class can be obtained simultaneously by these PHD filters; finally, by integrating TBM and the track particle labeling algorithm, the fusion result of multiple sensors is acquired. Simulations indicate that the performance of the proposed model-class-matched PHD filter is much better than that of the traditional PHD filter, and in the meantime the TBM framework of multiple sensors also improves the performance the proposed algorithm.