Feature-aided tracking in dense clutter using the Clutter09 data set

A feature-aided multihypothesis tracker (MHT) was used to process bistatic sonar data from the Clutter09 sea trial. The trial used a towed echo repeater that modified and retransmitted several variations of an incident signal including one that was intended to mimic interaction with a hollow steel cylinder. An automated aural classifier evaluated each of the contact signals in terms of its time-frequency and spectral characteristics, producing a set of feature values for each contact. The tracker used a subset of the feature information in its track score calculation to encourage the association of contacts that were similar in nature to the specified target. Comparisons of the baseline and feature-aided cases in three scenarios showed that the use of feature information in the tracker resulted in a factor of six reduction in the number of false tracks and a one-third reduction in the duration of the false tracks. There was limited impact on the quality of the specified target track. Tracks on other known objects, including the Campo Vega oil rig, were eliminated or degraded to the degree that they were dissimilar to the specific target.