The application of uniform design to particle swarm optimization for GK model parameter estimation

Parameter estimation for clutter model is a very important research in clutter modeling, simulation, classification and recognition. As a precondition for generalized K distribution(GK) clutter model, the speckle component and the amplitude modulated component should obey the generalized Gamma distribution, Parameter estimation for Generalized K distribution clutter model has high dimensional, nonlinear and other issues. This paper researches the parameter estimation for GK model based on improved particle swarm optimization. This method initialize particle swarm by uniform design method, can estimate the clutter model parameters effectively, is not sensitive to the initial value. Comparing with approximate maximum likelihood estimation and the basic PSO method, this method does not require iteration estimation, and reduces the computational complexity. The simulation results show that the method has good adaptability and accuracy, and verifies its validity and accuracy.