1: \begin{abstract} 

2: 

3: This paper investigates regularized estimation of Kronecker-structured covariance matrices (CM)  for  polarization radar in sea clutter scenarios where the data are assumed to follow the complex, elliptically symmetric (CES) distributions with a Kronecker-structured CM.

4: %The complex elliptically symmetric (CES) distributions are leveraged to  model the sea clutter.

5: To obtain a well-conditioned estimate of the CM, we add penalty terms of Kullback-Leibler divergence to the negative log-likelihood function of the associated complex angular Gaussian (CAG) distribution. This is shown to be equivalent to regularizing Tyler's fixed-point equations by shrinkage. A sufficient condition that the solution exists is discussed. An iterative algorithm is applied to solve the resulting fixed-point iterations and its convergence is proved. 

6: In order to solve the critical problem of  tuning the shrinkage factors, we then introduce two methods by exploiting oracle approximating shrinkage (OAS) and cross-validation (CV). 

7: The proposed estimator, referred to as the robust shrinkage Kronecker estimator (RSKE),  is shown to achieve better performance compared with several existing methods when the training samples are limited. 

8: %applied to the CM estimation for %the 

9: %polarization target detection \tj{with %in the context of measured 

10: %heterogeneous sea clutter, where the clutter} can be modeled using CES distributions with a Kronecker-structured CM. 

11: %When the training samples are limited, the proposed estimator, referred to as the robust shrinkage Kronecker estimator (RSKE), has better performance compared with several existing methods.  

12: Simulations are conducted for validating the RSKE %proposed estimator 

13: and demonstrating its high performance by using the IPIX 1998 real sea data. 

14: \end{abstract}

15: