Gaussian mixture implementation of PHD filter based on Dirichlet distribution

A Gaussian mixture implementation based on Dirichlet distribution is proposed for probability hypothesis density filter. Maximum likelihood criterion is selected for the estimation of parameters of mixture components. Dirichlet distribution is adopted as the prior distribution of mixing weights of Gaussian mixture components. The competitive nature among the elements in Dirichlet distribution is applied in driving the irrelevant components to extinction during the iteration procedure. The Gaussian mixture component pruning is implemented by this way. Simulation results show that the component pruning algorithm based on Dirichlet distribution is slight superior to the threshold algorithm in Gaussian mixture implementation of probability hypothesis density filter.