Automatic Target Recognition Classification System Evaluation Methodology

This dissertation research makes contributions towards the evaluation of developing Automatic Target Recognition (ATR) technologies through the application of decision analysis (DA) techniques. ATR technology development decisions should rely not only on the measures of performance (MOPs) associated with a given ATR classification system (CS), but also on the expected measures of effectiveness (MOEs) of the potential end product. The purpose of this research is to improve the decision-making process for ATR technology development. The basis of the research is a decision analysis framework that allows decision-makers in the ATR community to synthesize the performance measures, costs, and characteristics of each ATR system with the preferences and values of both the ATR evaluators and the warfighters.