Environmentally Adaptable Detector/Classifier

Environmentally robust detection and classification aids are critical to the ASW search mission in order to assist in decision making and reduce operator workload. In future multi-static, multi-receiver search missions, the number of alerts is expected to increase dramatically. An operator will simply not have the time to manually analyze and classify all the target and clutter returns without an algorithm to assist him. The goal of this SBIR effort is to develop environmentally robust algorithms that significantly increase detection and the ability to discriminate target returns from clutter, and ultimately reduce the operator workload. Many past efforts have been all-or-nothing ventures. Either the operator does it all by himself, or the algorithm completely replaces him in his effort, or at best has no interaction with him. A major problem with previous automated systems is that when the classifiers are exposed to a new and unknown environment, they suffer a drastic performance degradation. The effort proposed here will capitalize on information extracted in situ from initial insonification of the search area. Techniques are proposed to identify those echo return features which best exploit the current mission operating environment. Operator tools are proposed which keep him in the loop, reduce classification workload, and allow him to process large amounts of detections simultaneously.