AUTOMATIC DETECTION OF MINERAL DEVELOPMENT SITES FROM MULTISPECTRAL IMAGERY

THIS PROPOSAL SEEKS TO ESTABLISH THE FEASIBILITYOF APPLYING NEWLY DEVELOPED MORPHOLOGY-BASED IMAGE SCREENING AND DETECTION ALGORITHMS TO THE MINERAL DEVELOPMENT SITE DETECTION PROBLEM IN PANCHROMATIC AND MULTISPECTRAL IMAGERY AND TO DESIGN AN OVERALL PROCEDURE FOR INTEGRATING THIS TECHNOLOGY INTO AN EFFECTIVE, PRACTICAL AND USER FRIENDLY IMAGE ANALYST TOOL. OUR UNIQUE MORPHOLOGICAL APPROACH ENABLES US TO EXPLOIT SHAPE, SIZE, SPATIAL STRUCTURE, TECTURE AND SPECTRAL SHAPE TO RAPIDLY DETECT PROBABLE INSTANCES OF MINERAL DEVELOPMENT SITES IN OVERHEAD IMAGERY. THE COMPUTATIONAL ADVANTAGES OF THIS APPROACH MAKE IT IDEALLY SUITED FOR HIGH-SPEED PROCESSING OF LARGE VOLUMES OF IMAGERY USING SPECIALIZED HARDWARE OR FINE-GRAINED PARALLEL COMPUTERS SUCH AS THE CONNECTION MACHINE. THIS NOVEL IMAGE SCREENING APPROACHWILL REDUCE THE AMOUNT OF IMAGERY THAT NEEDS TO BE VIEWED AND EXPLOITED BY IMAGE ANALYSTS, PERMITTING THE ANALYSTS TO PERFORM WHAT THEY DO BEST, IMAGE INTERPRETATION RATHER THAN WIDE AREA SEARCH. OUR PHASE I PROPOSAL IS TO ESTABLISH FEASIBILITY OF OUR APPROACH FOR THE MINERAL DEVELOPMENT SITE DETECTION PROBLEM. EFFICIENT ALGORITHM IMPLEMENTATIONS ON A PARALLEL COMPUTER WILL OCCUR IN A PHASE II PROTOTYPING EFFORT. ANTICIPATED BENEFITS/POTENTIAL COMMERCIAL APPLICATIONS - THE PHASE IRESEARCH WILL DEMONSTRATE THE FEASIBILITY AND POTENTIAL OF THE MORPHOLOGICAL PROCESSING APPROACH AND LAY THE GROUNDWORK FOR FURTHER ALGORITHM VALIDATION AND PROTOTPYE DEVELOPMENT OF AN IMAGE ANALYST TOOL IN A PHASE II PROGRAM. THIS TECHNOLOGY IS APPLICABLE TO A NUMBER OF LABOR INTENSIVE TASKS IN IMAGE EXPLOITATION WHERE WIDE AREA SEARCH IS INVOLVED