Target Discrimination Techniques for Infrared Search and Track

Scanning shipboard IRST systems generate an unacceptable number of false alarms when viewing non-gaussian" backgrounds generated from sea reflectance, birds. etc. ISI proposes a major extension of the traditional IRST discrimination approach using a combination of likelihood ratio tests, neural networks, expert systems and fuzzy logic. ISI proposes to explore multiple implementation issues through the use of an interactive, rapid prototyping environment for building target discrimination systems (initially one conventional and one based on NN) The primary goal of the track discrimination system is to increase target detection while significantly reducing pfa. In Phase II, ISI will deliver a prototype system capable of a real-time demonstration when connected to a scanning shipboard IRST system as well the transition to a commercial product. Confidence in ISI's ability to achieve these goals is inspired by: ISI's involvement in the development of discrimination algorithms for fielded IRST systems, 0 Our R&D tools for deploying algorithms on parallel processors, 0 ISI's access to INTEL and Mercury parallel processors, 0 The successful completion of several Phase II SBIRs in related areas * A successful commercialization of DOD technology which now generates over one million dollars in annual sales.