Maximum Likelihood Estimation of K-distribution Parameters Using Number Theoretic Methods

摘要

The K-distribution is widely applied in synthetic aperture radar (SAR) image processing. However, the multi peak complicated likelihood function causes much trouble to obtain the maximum likelihood estimation of Kdistribution parameters. Based on the number-theoretic net (NTnet), the computable steps of sequential number-theoretic method for optimization (SNTO) were proposed to get the MLE of the parameters of K-distribution. Comparing with the nonML estimator Y0.1, we do Monte Carlo trails with different values of shape parameter and different sample sizes. The simulation results show that the proposed method outperforms the fractional moment based technique.