Recursive temporal-spatial information fusion with applications to target identification

Centralized/distributed recursive algorithms for temporal-spatial information integration using the Dempster-Shafer technique are developed. Compared with the Bayesian approach, the Dempster-Shafer technique has the strong capability of handling information uncertainties, which are particularly desirable in many applications. In the centralized integration algorithm, all information is pooled into the central processor and integrated. In contrast, the distributed integration algorithm shares the computational burden among the local processors, which increases the computational efficiency. The developed algorithms are effectively applied to a target identification problem with three sensors: identification of friend-foe-neutral (IFFN), electronic support measurement (ESM), and infrared search and track (IRST)