New Algorithms and Sparse Regularization for Synthetic Aperture Radar Imaging.

摘要

The PI led a collaborative effort to quantify the super-resolution potential of different computational methods for the directionfinding problem in sensing and surveillance. The difficulty of super-resolution is summarized in three quantities (the super-resolution factor, the signal-to-noise ratio, and the number of targets), and tight scalings between these quantities are presented to decide whether some methods can succeed -- or every method must fail -- at the target detection task. The analysis identifies the algorithms that perform well, and those that don't, even in the case of targets that shadow each other (nearby azimuths, different ranges).