In view of the low precision of particle filter algorithm and particle degradation in target tracking, a GH-RPF algorithm is put forward. Based on particle filter, Gauss-Hermite filter is applied to generate the importance density function, and meanwhile canonical transformation is adopted to re-sampling in order to improve the diversity of particles. If the algorithm is applied to nonlinear and non-Gaussian target tracking, it can be seen from the simulation result that the filtering accuracy is higher and tracking performance is better compared to the standard particle filter algorithm as well as EKPF.