Multiple Frame Cluster Tracking

"The aim of the BMD program is to provide a layered defense against ballistic missile threats.A central component to the BMD system is the tracking system which hasto provide accurate tracks for discrimination and prediction of thethreat's position. Countermeasures in the form of debris, chaff, spentfuel, and balloons can overwhelm any tracking system that tracksonly individual objects. Thus, tracking these groups of objects orclusters followed by transitions to individual object tracking if andwhen individual objects separate from the clusters is a necessarycapability for a robust and real-time tracking system.The objective of the proposed program is to demonstrate feasibility ofmultiple-frame cluster tracking capability for missile defense. Theproposed approach will form multiple clustering hypotheses on each frameof data and base individual frame clustering decisions on theinformation from multiple frames of data in much the same way that MFAor MHT work for individual object tracking. The combination thenpresents a seamless set of algorithms for cluster and individual objecttracking and transitions between the two for BMD. Although appropriateclustering methods are to be investigated, two candidate methods areBayesian EM clustering and the distance based k-means algorithm. Numerica, Inc. has a very successful record of developing algorithms(specifically in the target tracking area), implementing the algorithmsusing highly-struc