Forming regularized maximum likelihood strip-map synthetic aperture radar images using the block RLS algorithm

The data matrix that is needed for forming the maximum likelihood (ML) image in strip-map synthetic aperture radar (SAR) has block structure. This structure allows for a recursive way of forming a regularized ML image using the block recursive least-squares (RLS) algorithm. The regularization serves three purposes: Properly chosen, it provides a more stable solution; it combats noise; and it allows the block RLS algorithm to be initialized. In this paper, it is shown that an optimal regularization parameter exists for this problem and that the solution to the regularized normal equations for the strip-map SAR model is solved by the block RLS algorithm. Simulated results are shown.