ROBOTIC VEHICLE VIDEO IMAGE TRANSMISSION

UNMANNED VEHICLES ARE DESIRABLE IN HOSTILE ENVIRONMENTS. GUIDANCE IS A DIFFICULT PROBLEM, PARTICULARLY IN ROUGH TERRAIN. VISION IS ESSENTIAL. AUTONOMOUS GUIDANCE IS NOT PRESENTLY VIABLE. REMOTE VIDEO GUIDANCE BY AN OPERATOR IS A PRACTICAL SOLUTION. VIDEO DATA RATES IMPLY VERY HIGH CARRIER FREQUENCIES (GIGAHERZ) WHICH IN TURN REQUIRE LINE OF SIGHT TRANSMISSION. MANEUVERING OPTIONS ARE SERIOUSLY HAMPERED. A THREE-FOLD APPROACH IS PROPOSED TO REDUCE VIDEO DATA RATES, AND HENCE TRANSMISSION FREQUENCIES, BY MORE THAN TWO ORDERS OF MAGNITUDE. THE FIRST STEP IS TO RESAMPLE IMAGE DATA USING LOG-POLAR COORDINATES. WE HAVE SHOWN THAT THIS YIELDS BETTER THAN AN ORDER OF MAGNITUDE REDUCTION IN PIXEL COUNT, FOR A GIVEN RESOLUTION IN THE CENTER OF THE FIELD OF VIEW. THE SECOND STEP IS TO EXTRACT ONLY SALIENT IMAGE FEATURES FOR TRANSMISSION. THE THIRD IS SIGNAL COMPRESSION BY PROVEN TECHNIQUES SUCH AS BLOCK TRUNCATION CODING. THIS STUDY WILL IMPLEMENT SOFTWARE ALGORITHMS FOR IMAGE REMAPPING, FEATURE EXTRACTION AND SIGNAL COMPRESSION. THESE WILL BE TESTED ON REAL IMAGERY IN NON-REAL TIME. THE BEST CANDIDATES WILL BE SELECTED FOR PRELIMINARY DESIGN OF REAL TIME HARDWARE TO BE CONSTRUCTED IN PHASE II.