A Gaussian Sum Approach to the Multi-Target Identification-Tracking Problem.

摘要

The joint problems of identification, tracking and prediction in a multi-target multi-sensor environment are considered. Measurements giving information about the location of each target are to be processed in order to identify the type and to estimate the present state of each target. There is no a priori information relating a given measurement to a particular target. Once a target is properly identified its impact point is to be predicted. The previously developed Gaussian sum approach is used. The a posteriori density of the state of the target giving rise to the latest measurement is derived and an inherently parallel implementation of the algorithm is indicated. (Author)