Explicit bipercentile parameter estimation of compound-Gaussian clutter with inverse gamma distributed texture

Compound-Gaussian model with inverse Gamma distributed texture is a powerful tool to characterise heavy-tailed sea clutter. The parameters of the model need to be estimated in advance or online to assist the optimum detection. Moment-based estimators are suitable for a limited range of shape parameter, the maximum likelihood (ML) estimator lacks an explicit expression and fast iterative solution, and the method of log-cumulants (MoLC) can work in all cases but is not good for large shape parameters. Moreover, their performance is sensitive to outliers in samples. Explicit bipercentile estimators are proposed to estimate the scale and shape parameters, which are computationally efficient and outliers robust. Without outliers in samples, the bipercentile estimators are comparable with the moment-based estimators and MoLC estimator in performance except when the number of samples is very small. With outliers in samples, the bipercentile estimators are much better than the moment-based and ML estimators. Moreover, the ability of the bipercentile estimators is verified by measured sea clutter data.