Abstract : The research funded under this grant focused on several key challenges arising in automatic target recognition (ATR) systems. The robust estimation of geometric features is a critical aspect of ATR systems and thus methods for robust boundary extraction and feature enhancement were developed based on both statistical modeling and compressed sensing. Another set of challenges was related to novel, non-conventional sensing geometries arising in modern layered-sensing systems. Traditional sensing has focused on single sensors and single aspects, e.g. conventional mono-static, narrow aspect SAR. But as new sensing paradigms are considered, new methods for image estimation and processing are needed. In response, novel, robust methods for wide-angle image formation and multi-static multi-sensor data fusion were developed, based on powerful sparsity constraints. In addition, recent methods from compressed sensing were applied to SAR imaging problems of interest to the Air Force to reduce sampling requirements and improve robustness. Finally, imaging of scenes with moving targets has become a problem of great interest to AFRL. In response new methods for the formation and treatment of scenes with motion were developed based on over complete dictionaries.