RADIOGRAPHIC IMAGE PREDICTION AND COMPUTER MODELING

ARMY ARMAMENT DEVICES ARE CONTROLLED FOR SAFETY AND OPTIMUM PERFORMANCE BY A COMPLEX FUZE MECHANISM. FUZES ARE DESIGNED BY COMPUTER AIDED DESIGN (CAD), FABRICATED BY HAND, AND TRANSPORTED SEPARATELY FROM THE EXPLOSIVE CHARGE OF THE ARMAMENT FOR SAFETY. EACH FUZE IS INSPECTED BY X-RADIOSCOPY. THE PRESENT RADIOSCOPIC METHOD IS EMBODIED IN THE AUTOMATIC FUZE INSPECTION BY RADIOGRAPHY SYSTEM (AFIRS), WHICH USES FEATURE RECOGNITION TECHNIQUES TO DETECT FUZE MISASSEMBLY FLAWS. THE REFERENCE PATTERNS FOR THE FEATURE RECOGNITION PROCESSOR COMPRISE ACTUAL RADIOSCOPIC IMAGES OF PROPERLY AND IMPROPERLY ASSEMBLED FUZES. STATISTICAL INFORMATION IS CALCULATED FROM THE IMAGES AND USED IN THE PROCESS. THE DATABASE IS DIFFICULT TO COMPILE SINCE FLAWS MUST BE HAND-FABRICATED INTO FUZES. THIS PROPOSAL IDENTIFIES THIS PROBLEM AND OFFERS A SOLUTION IN RADIOSCOPIC IMAGE MODELING. MODELLED IMAGES INTRODUCE IMPROVED DATABASE COMPILATION SINCE THERE IS NO HAND-FABRICATION INVOLVED AND THUS THE DATABASE CAN BE LARGER AND MORE REPRESENTATIVE OF THE FUZE COMPONENTS AND ASSOCIATED FLAWS. PHASE I WORK WILL AIM TO DEVELOP A CAD GRAPHICS FUZE DESCRIPTION WHICH ACCOMPLISHES SIMULATED INSPECTION BY INTERFACE TO SOURCE MODELING SOFTWARE, OBJECT ATTENUATION DATA, AND DETECTOR RESPONSE CHARACTERISTICS. IN PHASE II, WE WILL FURTHER DEVELOP THE RADIATION MODEL AND INTERFACE TO AFIRS TO DEMONSTRATE IMPROVED FUZE INSPECTION CAPABILITY.