Monotonic Iterative Algorithms for SAR Image Restoration

We present an iterative algorithm for restoring synthetic aperture radar (SAR) images with arbitrary gaps in spatial frequency. In some SAR applications a contiguous frequency band is not recorded which results in gaps in the SAR image spectrum. These gaps manifest themselves as artifacts which degrade the quality of the associated SAR image. Maximum a posteriori (MAP) techniques allow one to compensate for these spectrum gaps but can result in a difficult high-dimensional optimization problem. By using concepts collectively known as optimization transfer or MM-algorithms, one can easily design efficient iterative algorithms that are tuned for a particular objective function and have provable convergence properties. In this paper we will review MAP-based SAR image restoration, derive an iterative algorithm that monotonically minimizes the associated objective function, and demonstrate it on actual SAR imagery.