A probabilistic-statistical approach to target tracking in a cluttered environment.

摘要

Track initiation, confirmation, termination, and maintenance in cluttered environments are difficult mainly due to the so-called measurement—origin uncertainty—a measurement could have been generated by a target, clutter or some other sources. Traditional estimation methods can not be applied directly. To solve the problem of measurement-origin uncertainty, this dissertation proposes two new probabilistic approaches based on perceivability and target-measurement-support (TMS); develops two new probabilistic data association (PDA) trackers based on perceivability, and a new PDA filter based on TMS; applies two statistical approaches of the most probable confirmed track (MPCT) and sequential probability ratio test (SPRT) for track confirmation and termination. This dissertation also develops several clutter density estimators and designs the tracker parameters for target tracking in a cluttered environment.