Hierarchical Multitarget Tracking and Classification - A Bayesian Approach

The tracking and classification of multiple targets by a network of local agents (nodes) is considered. A Bayesian approach is adopted as the theoretical basis. Each local agent processes the local sensor data to obtain the local information state consisting of the local hypotheses, tracks and their relevant probabilities and state distributions. These are communicated to the fusion agent (node) who tries to reconstruct the global information state conditioned on the data which would be available if they were communicated from the local agents. Both results for static and dynamic target models are presented assuming feedback from the fusion agent.