A grouped weighted covariance matrix estimator in partially homogeneous sea clutter

A covariance matrix estimator in spatially partially homogeneous sea clutter is addressed in this paper. The partial homogeneity of sea clutter and the unavoidable false targets lead to a large error when the covariance matrix of sea clutter is estimated by the traditional estimators. In order to reduce the estimation error, the secondary data of sea clutter are grouped, the maximum posteriori estimation of texture is used as a weighted coefficient in each group, and a grouped weighted sample covariance matrix (GW-SCM) estimator is proposed. Due to the existing false targets, a consistent factor measuring the difference between covariance matrices is proposed. The group where interference lies in is identified by the proposed parameter, and the interference is eliminated at last. The experimental results of the real data show that the proposed grouped covariance matrix estimator not only eliminates false targets effectively, but also outperforms the estimator without grouping about 3dB in the partially homogeneous sea clutter.