All-Weather Feature-Based Combat Identification

Developers of ATR systems for SAR imagery have begun exploring the use of features to reduce the computational complexity of pixel based pattern matching approaches, and to improve ATR performance in challenging target regimes. In addition, model based ATR approaches are emerging as a viable means of coping with the combinatorics of real world operating conditions. Rather than rely on massive databases of pre-stored, and often incomplete, signature exemplars, these approaches use target models and computational electromagnetic code to predict target signatures "on the fly." This effort will investigate the impact on model based ATR performance of various topographic SAR features. Experiments will be conducted under a range of realistic F-18 sensing, environment, and target conditions related to Combat ID missions. Optionally, the effort will design and deterministically assess the real-time processing performance of an F-18 model based SAR ATR system for implementation in Phase II.