Model-based integrated HRR object tracking and classification

Radar target classification based on 2D stochastic object model matching is studied in this paper. A network of high range resolution (HRR) radars provides range measurements at multiple time steps, while the extended object is moving in the surveillance area. Alignment of the multi-aspect HRR data in a common 2D coordinate system is required. For this reason, tracking of the extended object is integrated in the classification algorithm. The novelty of this work is the estimation of the object state, conditioned on the object class, by applying the HRR data on a particle filter. A coarse-to-fine multi-resolution data processing scheme is introduced in the filter update step. The reason is the experienced filter degeneracy, when working with HRR data. The proposed processing scheme achieves accurate auto-alignment of the HRR data. The classification system successfully distinguishes between very similar objects, by applying 2D object model matching.