SUBOPTIMAL TECHNIQUES FOR TRACK-ORIENTED MULTIPLE HYPOTHESIS TRACKINGALGORITHM AND JPDA ALGORITHM FOR MULTITARGET TRACKING TO BE EQUIVALENT

As algorithms to simultaneously track multiple targets in a high-densi ty environment where other targets or clutter exist near the tracking target, the MHT (Multiple Hypothesis Tracking) and JPDA (Joint Probabi lity Data Association) methods have drawn recent attention. The MHT is a multiple hypothesis correlation method with track initiation, track maintenance, and track deletion functions. The JPDA is practical beca use target state estimates can be uniquely computed at each sample tim e, although it has only the track maintenance function. Further, the t rack-oriented MHT is an improvement over the conventional MHT, with an improved tracking initiation function. However, the track maintenance capability is not as yet clear. In this paper, a suboptimal technique is found in order to allow the track-oriented MHT to have the same al gorithm as the JPDA. This method combines several hypotheses into one hypothesis, consisting only of the previously confirmed track, namely, the track for which the process of track initiation is already comple ted. In addition, target state estimates such as position and speed of the confirmed track are computed by using the reliability of the hypo thesis. As a result, it is found that the JPDA is a special case of tr ack-oriented MHT. The design of a track-oriented MHT with an excellent track maintenance property of the JPDA becomes easier. (C) 1998 Scrip ta Technica.