Abstract:

In the modeling, simulation and classification of the clutter, the estimation of model parameters of the clutter is an important research area. For the frequently adopted generalized K-distribution clutter model, the speckle and amplitude modulation components are both assumed to obey the generalized Gamma distribution. It turns out that the parameter estimation in this model is difficult due to high-dimensionality and nonlinearity. In order to solve this problem, this paper applies the improved particle swarm optimization (PSO) to the estimation of parameters of the generalized K-distribution. Specifically, the paper adopts the uniform design method to initialize the particle swarm and employs the strategy of across and mutation to improve the global convergence performance of the standard PSO. In fact, the proposed method can accurately estimate each parameter of the clutter model. Moreover, the method has the advantages of low computation burden, fast convergence rate and preferable stability. It is demonstrated by simulation results that the method is of good adaptability and estimation accuracy, which proves its effectiveness and exactness.