Multitarget Tracking of Distributed Targets Using Histogram-PMHT

It has been shown previously that the Histogram Probabilistic Multi-Hypothesis Tracking (HPMHT) algorithm enables tracking of targets distributed across several sensor cells. Two related aspects of the H-PMHT algorithm are discussed in this paper, namely the estimation of target spread, and the use of the negative multinomial distribution to compensate for unobserved sensor cells. The examples presented show how the negative multinomial reduces bias in the target spread estimates near the edges of the sensor display. Examples are given in the context of multi-target tracking on a sonar broadband display.