Model-based Automated Target ID and Pose Estimation from SAR and ISAR Data

Automatic Target Recognition and intent determination from airborne Synthetic Aperture Radar (SAR) or Inverse Synthetic Aperture Radar (ISAR) imaging requires highly accurate pose determination. Prior knowledge in the form of radar signatures of 3D target models that accurately capture locations of a few (<10) major features within a target, in conjunction with 3D pattern matching to GPS positioning and high bandwidth SAR/ISAR imaging, may provide a means to achieve highly accurate pose determination. In Phase I, the Augmented QUID Synthetic Aperture Radar (QUASAR) code was developed to test the hypothesis using collections of spheres and cylinders to represent 3D targets. In Phase II, key elements of QUASAR will be extended for more accurate image reconstruction and target model simulation: the simulation of range-cross range-height images, 3D pose estimation using a database search of pre-computed SAR and ISAR images, and radar cross section feature modeling of flat plates and dielectric and semi-conducting materials. QUASAR image reconstruction will be generalized to two 3D SAR imaging configurations: sensors with combined elevation and azimuth apertures and multiple sensor interferometric SAR. The code will be validated with a highly accurate database of ISAR satellite images generated using Xpatch.