AUTOMATIC DETECTION OF POWERLINES FROM SAR AND MULTISPECTRAL IMAGERY

THIS PROPOSAL SEEKS TO ESTABLISH THE FEASIBILITY OF APPLYING NEWLY DEVELOPED MORPHOLOGY-BASED DETECTION ALGORITHMS TO THE POWERLINE DETECTION PROBLEM IN SAR AND MULTISPECTRAL IMAGERY AND TO DESIGN AN OVERALL PROGRAM TO INTEGRATE THIS TECHNOLOGY INTO A COMPLETE MAP VALIDATION AND UPDATING SYSTEM. THE IMPORTANCE OF THIS OUR APPROACH IS TWOFOLD. FIRST, DETECTION PERFORMANCE CAN BE DRAMATICALLY IMPROVED FOR THE DETECTION OF TARGETS IN SEVERE BACKGROUND CLUTTER, IN WHICH EXCESSIVE FALSE ALARMS ARE A TYPICAL PROBLEM. SECONDLY, MORPHOLOGY-BASED ALGORITHMS CAN BE IMPLEMENTED IN VERY HIGH-SPEED HARDWARE BECAUSE THE FUNDAMENTAL OPERATIONS INVOLVE ONLY LOGICAL COMPARISONS (MIN AND MAX). THE ADVANTAGES OF THIS APPROACH MAKE IT IDEALLY SUITED FOR PROCESSING LARGE VOLUMES OF SAR AND MULTISPECTRAL DATA AT VERY LOW COMPUTATIONAL COST USING SPECIALIZED HARDWARE OR FINE-GRAINED PARALLEL COMPUTERS. OVER THE PAST SEVERAL YEARS ATLANTIC AEROSPACE ELECTRONICS CORPORATION HAS DEVELOPED AND DEMONSTRATED THIS MORPHOLOGY-BASED TECHNOLOGY FOR DETECTING NON-RESOLVED AND RESOLVED TARGETS IN HIGH CLUTTER BACKGROUNDS. THIS WORK WAS ACCOMPLISHED IN SEVERAL PROGRAMS SPONSORED BY THE AIR FORCE, IN WHICH BASIC ALGORITHM COMPONENTS WERE DEVELOPED, ALGORITHM PERFORMANCE WAS QUANTIFIED ANALYTICALLY AND THEIR OPERATION WAS VERIFIED EXPERIMENTALLY USING RECORDED REAL DATA FROM AIRBORNE SAR AND GROUND-BASED DOPPLER RADARS, AS WELL AS PASSIVE ELECTRO-OPTICAL SENSORS. OUR PROPOSAL IS TO EVOLVE THIS EXISTING TECHNOLOGY TO THE POWERLINE DETECTION PROBLEM. ANTICIPATED BENEFITS/POTENTIAL COMMERCIAL APPLICATIONS - THE PHASE I RESEARCH WILL DEMONSTRATE THE FEASIBILITY AND POTENTIAL OF THE MORPHOLOGICAL PROCESSING APPROACH AND LAY THE GROUNDWORK FOR FURTHER ALGORITHM DEVELOPMENT IN A PHASE II PROGRAM USING EXTENSIVE RECORDED SENSOR DATA. THIS RESEARCH WILL ALSO RESULT IN THE DEFINITION OF A LONG TERM PROGRAM TO INTEGRATE THIS TECHNOLOGY INTO MAP UPDATING SYSTEMS.