Estimators of compound Gaussian clutter with log-normal texture

ABSTRACT

Estimators of clutter models parameters based upon higher order moments estimator (HOME) produce usually poor results in particular for low sample sizes. In an attempt to remedy this situation, closed forms of [zlog(z)] and fractional order moments estimator (FOME) are derived in this work and yield a good estimation accuracy of parameters of the compound Gaussian clutter with log-normal texture (CG-LNT). Using simulated and real data, estimation comparisons show that best values of mean square error (MSE) and bias are achieved using the proposed procedures.