Model-Based Optimization Using l_1 Minimization for Reducing the Uncertainty in Radar Cross-Section (RCS) Measurements and Predictions

摘要

Our team at Integrity Applications Inc. (IAI) has worked extensively on radar cross-section (RCS) measurements and predictions and problems related to removal of background contamination, defect detection and localization, image editing and reconstruction (IER), sub-Nyquist interpolation, and near field-to-far field transformation (among others). In all cases, we have found that model-based optimization techniques using an l_1 minimization solver provide significantly improved performance with forward models as simple as isotropic point scatterers and as complex as rigorous method of moments (MoM) codes. In this overview paper, we present two simulated examples of model-based optimization using l_1 minimization from each end of that spectrum. The first involves the use of a sparsely-sampled set of RCS measurements to reduce the uncertainty in a MoM model due to unknown defects on the target. The model is then used to interpolate the sparsely-sampled RCS pattern data. The second involves the use of a dictionary of point scatterers and other linear basis functions to reduce the uncertainty in a set of RCS measurements due to additive contamination from clutter and noise.