Performance assessment for geometric-based covariance estimation in heterogeneous clutter

We consider the problem of covariance matrix estimation in heterogeneous environments for radar signal processing applications, where the secondary data exhibit heterogeneity in local power and share the same covariance structure. Without any knowledge of the statistical characterization of the sample support, a class of estimators, delined as the geometric barycenters of some basic covariance matrix estimates obtained from the available secondary data set, are proposed by exploiting the characteristics of the positive-delinite matrix space. Finally, we evaluate the performance in terms of detection properties of an adaptive matched lilter (ANMF) with the proposed estimators in the presence of compound-Gaussian disturbance.