Artificial Neural Network Tracking of Targets in Nuclear Backgrounds

WE WILL ADAPT AND IMPROVE AN INNOVATIVE MOVING-TARGET DETECTION AND TRACKING APPROACH, BASED ON ARTIFICIAL NEURAL NETWORK (ANN) PROCESSING TECHNIQUES, FOR APPLICATION TO THE TRACKING OF BOOST-PHASE MISSILE TARGETS IN THE PRESENCE OF INFRARED CLUTTER DUE TO HIGH-ALTITUDE NUCLEAR EXPLOSIONS (BANES). OUR RESEARCH WILL USE TIME-SEQUENCED SIMULATED INFRARED IMAGES OF BANE BACKGROUNDS WITH MOVING TARGETS. THESE IMAGES WILL BE PROVIDED BY PHOTON RESEARCH ASSOCIATES (PRA) USING THE SDIO/NRL STRATEGIC SCENE GENERATION MODEL (SSGM). ALGORITHM TRAINING AND PERFORMANCE TESTING WILL BE DONE FOR REPRESENTATIVE LEVELS OF TARGET SIGNAL-TO-NOISE RATIO, AND NUCLEAR CLUTTER LEVEL, AND RESULTS WILL BE PRESENTED IN THE FORM OF PROBABILITY OF DETECTION VERSUS PROBABILITY OF FALSE ALARM CURVES. THE TRACKING ALGORITHM WILL INCLUDE STANDARD NUCLEAR-CLUTTER SUPPRESSION TECHNIQUES AS A PRE-PROCESSING STEP.