A Clutter Reduction Technique for Microwave Reconstruction of Shallow Underground Targets

This paper presents a signal pre-processing technique for the stepped-frequency ground penetrating radar that is aimed at suppressing clutter related to antenna effects and variation of the ground surface response. Both the ground bounce and the clutter can obscure shallow underground objects such as landmines making difficult their detection. At the preprocessing stage, the measurement system has been calibrated using a free space and metal sheet as standards. Both the modulus and the phase of the signal obtained for metal sheet have been approximated by polynomials, and this synthetic signal is used to derive approximate transfer functions of the antenna. This technique utilizes calibration measurement for only one position of the metal sheet and yet reduces clutter related to the antenna effects. Besides, the calibration simplifies further subtraction of the ground surface response. The 2-D image of the subsurface is built then using B-scan data from which the ground bounce is eliminated. The performance of the method is demonstrated in several examples employing the frequency-domain data collected by a transmitting-receiving antenna pair.