REAL-TIME IMAGE PRE-PROCESSOR

IN REAL-TIME TRACKING AND POST-FLIGHT VIDEOTAPE DATA EXTRACTION APPLICATIONS AN IMPORTANT REQUIREMENT EXISTS FOR RECONSTRUCTING AND ENHANCING IMAGES IN REAL-TIME (60-FIELDS PER SECOND). VARIOUS FILTERING TECHNIQUES MUST BE APPLIED TO THE IMAGES FOR NOISE REMOVAL, EDGE ENHANCEMENT OR OTHER "ORE-PROCESSOR" FUNCTIONS. THESE RECONSTRUCTED AND ENHANCED IMAGES WILL BE USED BY AN AUTOMATIC TRACKING SYSTEM EMPLOYED FOR MISSILE TEST RANGE INSTRUMENTATION. IT IS PROPOSED THAT AN ENTIRELY NEW PRE-PROCESSOR DESIGN BE INVESTIGATED, BASED UPON THE USE OF ARRAYS OF CUSTOM INTEGRATED CIRCUITS TO PROCESS THE IMAGE AND A DEDICATED SPECIAL-PURPOSE PARALLEL PROCESSING ARCHITECTURE. THIS DESIGN CAN PROVIDE RECONSTRUCTION BY MEDIAN FILTERING AS WELL AS ANY DESIRED CONVOLUTION OPERATION (SUCH AS EDGE ENHANCEMENT OR SEGMENTATION), IN REAL TIME. BY DESIGNING IN THIS FLEXIBILITY FROM THE BEGINNING, ANY NEW ALGORITHMS DEVELOPED CAN BE IMMEDI-ATELY IMPLEMENTED WITHOUT ANY HARDWARE CHANGES REQUIRED. CURRENT PRE-PROCESSORS THAT CAN DO IMAGE ENHANCEMENT OPERATIONS AT 60 FIELDS PER SECOND ARE BASED UPON SOFTWARE ALGORITHM IMPLEMENTATIONS AND ARE ABLE TO DO ONLY A SMALL SUBSET OF THE VARIOUS TYPES OF FILTERING THAT MIGHT BE DESIRED AT HIGH SPEEDS.