Application of entropy classification method to the detection of subsurface linnear targets in polarimetric GPR data

Although fully polarimetric analysis techniques have been applied to remote sensing radar for charactering different surface scattering properties using, these techniques have not yet been widely adopted for Ground Penetrating Radar (GPR) applications. In 2015 we applied the “H-alpha decomposition” technique to polarimetric GPR data for classifying buried metallic targets such as wire branches, a sphere, a plate, and a dihedral. The H-alpha decomposition can separate scatters into four different types according to different depolarization effects: linear polarization, random depolarization (branches, roots, random media), no depolarization (sphere, plates, and horizontal layers), and 90 degrees depolarization (corner reflectors). In this paper, we explore the utilization of H-alpha decomposition method in clutter reduction and linear target detection during the preprocessing stage so that automatic linear target detection can be achieved in poor signal-to-clutter ratio (SCR) and signal-to-noise ratio (SNR) environments.