Multi-Target Joint Tracking And Classification Based On Mmphd Filter And Tbm Framework

To solve the multi-target tracking and classification problem in clutter measurements, this paper introduces a recursive algorithm, which is based on the multiple model probability hypothesis density (MMPHD) and transferable belief model (TBM) framework using multiple kinematic radars with the particle implementation. Considering joint tracking and classification (JTC) simultaneously has been an essential problem, our proposed algorithm adopts TBM and prior information instead of the feature measurements to classify the targets. In the prediction stage, the particles are propagated according to their class-dependent model in PHD filters with class label. Then, the measurements from different sensors update their particle weight. The particles and their corresponding weights represent the estimated PHD distribution in different sensors. These PHD distributions are used to jointly estimate their states and class. Finally, using the TBM framework and target labelling techniques integrate targets state and class probability in various sensors. Simulation results are presented to show the effectiveness of our proposed algorithm over the traditional MMPHD and indicate our multi-sensor algorithm based on TBM framework is much better than the single sensor algorithm in all respects.