Five-Parameter Compound-Gaussian Model for Dual-polarimetric Sea Clutter with Bivariate Lognormal Texture

In maritime radar systems, dual-polarization significantly enhances target detection and recognition capabilities. Dual-polarimetric sea clutter data exhibit strong polarimetric correlation in power. Nowadays, no model can characterize polarimetric correlation in dual-polarimetric sea clutter. This paper proposes a five-parameter compound-Gaussian model with bivariate lognormal texture to effectively model dual-polarimetric moderate/high-resolution sea clutter. The five parameters in the bivariate lognormal distribution reflect the non-Gaussianity and powers of sea clutter at individual polarizations and inter-polarimetric texture correlation. It is known as the compound-Gaussian model with bivariate lognormal texture (CGBLN). Further, the parameter estimation of the CGBLN is investigated and several estimators are given to estimate the texture polarimetric correlation coefficient. Experiments with measured dual-polarimetric sea clutter datasets validate the CGBLN model, demonstrating its satisfactory fit to the measured sea clutter data. Moreover, the parameter estimators are also verified to be effective and robust to outliers. The proposed dual-polarimetric sea clutter model provides the foundation to develop better coherent detection methods for dual-polarimetric maritime radars.