Iterative maximum likelihood estimation of the compound inverse Gaussian clutter parameters

Abstract

This Letter aims at proposing a cost-effective method for estimating the parameters of the compound inverse Gaussian distributed clutter. Under the assumption of the absence of thermal noise, an iterative maximum likelihood estimator (IMLE) is proposed and compared with existing MLE, the [zlog(z)] estimator, the non-integer order moments estimator and the higher order moment estimator. The results obtained show that the IMLE method outperforms all the other methods and has similar estimation performance than the MLE method but requires lower computational time.

Self-learning parameter estimation of K-distributed clutter using nonlinear GBDT model

“…In the compound Gaussian model (CGM) [1][2][3][4], sea clutter time series is modelled by the product of a slowingvarying texture and a fast-varying speckle , i. e. = √ . When the texture follows Gamma distribution, the amplitude of sea clutter = | | follows the famous K distribution and its PDF and CDF are ( ; , ) = ( )…”

“…Introduction: It is meaningful and important for marine radars to investigate the amplitude statistical model of sea clutter in detection and tracking [1][2][3]. K distribution is a widely-used and effective amplitude probability model of sea clutter at low and moderate range resolution [4].…”