Tracking and identification for closely spaced objects in clutter

The sampling based bootstrap filter is applied to a measurement association and classification problem for two adjacent objects which gradually separate. The problem is to apply position and discrimination (signature) information to identify and track the objects. Due to the object proximity and the presence of dense clutter, the association between measurement/classification data and the objects is initially highly uncertain. The bootstrap filter is employed to integrate the available information in near-optimal fashion without recourse to complex hypothesis formulation. Thus the posterior distribution of the two objects is generated.