Knowledge-Based Target Recognition

Target recognition systems have undergone a variety of changes over the last 20 years. Initial systems focused on exploiting signal processing techniques to detect ground-based targets based on one dimensional signals. Limitations of these systems eventually led to the development of advanced target recognizers (ATR) that processed two dimensional digital images to detect, classify, and identify targets. Though their perfonnance exceeded signal processing systems, ATRs exhibited several deficiencies to which artificial intelligence (AI) offered numerous potential solutions. The paper reviews the evolution of target recognition systems with primary focus on artificial intelligence applications. Deficiencies of AI approaches to target recognition are presented and complemented by a discussion of a blackboard-based ATR system currently being developed at Georgia Tech.