REMOTE DETECTION OF MOBILE MISSILE LAUNCHERS

TERA RESEARCH WILL INVESTIGATE THE USE OF FRACTAL SIGNAL PROCESSING TO DETECT MOBILE MISSILE LAUNCHERS IN REMOTLY SENSED IMAGERY. THIS NEW APPROACH BASED ON SELF-SIMILAR STOCHASTIC MODELS PROVIDES DIRECT DATA FUSION OF MULTISPECTRAL AND SAR IMAGERY. INHERENT PARALLESLISMS IN THE ALGORIGHM ALLOWS REAL TIME IMPLEMENTATIONS WITH PARALLEL PROCESSOR ARCHITECURES. WITH THIS APPROACH, MISSILE TARGETS ARE REPRESENTED BY A FRACTAL TIME SERIES MODEL CORRESPONDING TO A L/F STOCHASTIC PROCESS. THE DIGITAL IMAGES ARE MODELED AS A SUM OF FRACTAL PROCESSES CORRESPONDING TO THE SUPERPOSITION OF TARGET, BACKGROUND AND NOISE. PARAMETERS OF THE MODEL INCLUDING FRACTAL DIMENSION WILL BE USED TO DETECT AND IDENTIFY TARGETS OF INTEREST. THIS APPROACH DIFFERS FROM PREVIOUS FRACTAL APPROACHES IN AT LEAST TWO WAYS. FIRST, THE FRACTAL ESTIMATOR IS OPTIMAL IN THE SENSE THAT IT IS BASED ON THE WAVELET TRANSFORM, WHICH IS A CANDIDATE WHITENING FILTER FOR THE 1/F PROCESS. THIS APPROACH THEREFORE ACCOUNTS FOR THE EFFECT OF NOISE ON THE ESTIMATION PROCESS. SECOND, THE ENTIRE FRACTAL PROCESS IS ESTIMATED, RATHER THAN JUST A SINGLE PARAMETER, SUCH AS THE FRACTAL DIMENSION. THEREFORE, THE REPRESENTATION IS MUCH RICHER AND WILL PERFORM MORE EFFECTIVELY IN THE PRESENCE OF NOISE.