IDENTIFICATION

THERE IS A NEED TO DEVELOP FAST, ('REAL-TIME'), THREE DIMENSIONAL (3D) VISION PROCESSING ALGORITHMS APPLICABLE TO MACHINE RECOGNITION OF PATTERNS IN SUPPORT OF FIRM CONTROL SUB-SYSTEMS, CAPABLE OF AUTOMATICALLY RECOGNIZING TARGETS ON THE BATTLEFIELD. TARGET IDENTIFICATION WHICH INVOLVES COMPLEX HUMAN DECISION MAKING IS NOT VERY RELIABLE UNDER THE STRESS OF COMBAT ENGAGEMENT, EVEN THOUGH A DISTINCT AND UNAMBIGUOUS TARGET IMAGE IS AVAILABLE. IN MANY PRACTICL APPLICATIONS OF AUTOMATED VISION, THE EXACT SHAPE OF THE OBJECT TO BE RECOGNIZED IS KNOWN IN ADVANCE. FOR THE U.S. ARMY, IN MOST INSTANCES, THE BATTLEFIELD TARGETS TO BE RECOGNIZED WILL BE COMBAT AND TRANSPORT VEHICLES, MISSILE AND GUN CARRIERS, AND INFANTRY SUPPORT HELICOPTERS AND AIRCRAFT. IN THIS KIND OF SITUATION, IT IS DESIRABLE AND POSSIBLE TO SPLIT THE COMPUTATION INTO TWO STATES: A MODEL-BUILDING STAGE IN WHICH USEFUL INFORMATION ABOUT THE UNCHANGING TARGET CHARACTERISTICS IS COMPILED, AND AN EXECUTION STAGE IN WHICH THIS INFORMATION IS EXPLOITED FOR THE RAPID RECOGNITION OF TARGETS OF INTEREST IN THE IMAGE. THE RESULT OF OUR PROPOSED RESEARCH WILL BE A FAST, MACHINE-VISION PROGRAM FOR RECOGNIZING AND LOCATING BATTLEFIELD TARGETS IN IMAGES WITHOUT RESTRICTION ON THE ORIENTATION OF THE TARGETS IN SPACE. SUCH AN ARTIFICAL INTELLIGENCE (AI)-BASED PROGRAM WITH A HEURISTIC DATABASE WILL BE DEVELOPED ON A 32-BIT MICROPROCESSOR (E.G., MICROVAX II) AND WOULD EVENTUALLY RESIDE INSIDE A COMBAT OR VERY INTELLIGENT SURVEILLANCE AND TARGET ACQUISITION (VISTA) VEHICLE.