Regularized polarimetric target detection in the presence of inhomogeneous and/or non-stationary clutter

The polarimetric information of radar targets can be exploited to improve the detection performance, especially when the spatial-Doppler processing is insufficient. In the case of inhomogeneous and/or non-stationary clutters, however, the potential of the polarimetric detector is restricted due to the limited training data size, particularly when the sample support is less than the number of array elements. This paper presents an iterative shrinkage based regularized polarimetric target detection scheme to deal with the mentioned under-sampling problem in the presence of either Gaussian or compound Gaussian clutters. The involved shrinkage parameter in the convex combined sample covariance matrix or fixed-point covariance matrix is herein adaptively determined by using a newly developed neighbor point matching technique. Performance assessment using synthetic and real data shows that the proposed detector outperforms the non-regularized counterpart in terms of the probability of detection.