A Partitioning Method for Gamma Gaussian inverse Wishart Probability Hypothesis Density Filter using Kolmogorov-Smirnov Test

In multiple extended objects tracking, objects produce more than one measurement per time step. When objects are spatially close or maneuvering, the performance of typical partitioning methods will be reduced. This paper presents a partitioning method for gamma Gaussian inverse Wishart Probability Hypothesis Density (GGIW-PHD) filter based on the hypothesis test and the EM algorithm. The Kolmogorov-Smirnov test is modified to evaluate normality of the measurements. It provides guidance to the EM algorithm. The proposed partitioning method for GGIW-PHD filter performs well when the extended objects are spatially closed, and do not be affected by the inaccurate predicted components. Simulation results show that the proposed method performs better than existed algorithms.