An inhomogeneous clutter-clustered estimation method of covariance matrix

This paper addresses the problem of covariance matric estimation for radar adaptive constant false alarm rate (CFAR) detection in clutter-dominated disturbance modeled as compound-Gaussian process. For estimation purposes we resort to range cells, free of signal components, can be clustered into groups of data with one and the same value of texture. Extending the homogeneous clutter-clustered estimator to a more generalized situation and an inhomogeneous clutter-clustered estimator (ICCE) is obtained. Furthermore, to improve the estimation accuracy a recursive ICCE (RICCE) is proposed. The ICCE and the RICCE both have the CFAR property with respect to the statistics of the texture and the clutter covariance matrix. Compared with the existing estimators, the simulation shows that the RICCE has the higher estimated accuracy.