Mixture reduction techniques for Multiple Hypothesis Tracking of targets in clutter

Two complementary mixture reduction algorithms for Multiple Hypothesis Tracking (MHT) are presented. The first approach (MHT-2) uses the Integral Squared Error (ISE) in a simple optimization process, where the major principal axis of the optimal one-component fit, is used to guide the search for a near-optimal two-component fit. The second, less rigorous, approach (MHT-PE) Prunes unlikely hypotheses then Eliminates duplicate components, using the normalized overlap integral. Both methods are compared with PDA and other MHT mixture reduction techniques in Monte Carlo simulations.