Automatic Target Recognition (ATR) for Combat Identification (CID) represents one of the most demanding disciplines in sensor technology facing the United States Air Force today. Central to this process is the development of validated synthetic. Unfortunately, the synthetic database generation process is hampered by uncertainty. Under a previous program, Analytic Designs, Incorporated developed a robust and extensible theoretical foundation uncertainty analysis for Physical Optics based Radar Cross Section (RCS) prediction. The goal of this Phase~I proposal is to extend this statistical theory to include two important areas of uncertainty: coupled regions and altered parameterizations. Coupled regions addresses the statistical dependence associated with vertices comprised by feature-based regions. During the target modeling process, a modeler may be unsure of the exact location of a specific feature (e.g., hatch, weapon, antenna, etc.), and this uncertainty may be larger than the geometry capture process. Altered parameterizations addresses the need to simulate RCS uncertainty in synthetic data due to errors in pose angle and/or frequency to facilitate comparisons to measured data and streamline the data validation process. The overall purpose of this Phase~I program is to improve the efficiency of the synthetic RCS database development process and thereby advance Automatic Target Recognition capabilities.
