Target tracking algorithm based on improved Gaussian mixture particle filter

A improved Gaussian mixture particle filter algorithm was proposed to overcome the sample depletion brought by resampling step in particle filter. The algorithm which based on the characteristics of SPKF and particle filter, used SPKF to update and generate the proposal distribution. Then combined with measurement of the important sampling, it used limited Gaussian mixture model to approximate the posterior density of states. Finally, the traditional process of particle filter resampling was replaced by the greedy expectation maximization (EM) algorithm. The effects caused by sampling depletion were lessened. It is demonstrated by computer simulation that GEM-GMPF outperforms the one based on PF and the one based on EM-GMPF in tracking accuracy, and stability. Therefore it is more suitable to the nonlinear state estimation.