Research on Effect of Antenna Beamwidth for Compressive Sensing Based GPR Imaging

Compressive sensing (CS) techniques have been exploited in subsurface imaging owing to its reduction in detection time and high-quality performance in imaging. Generally, the received data is represented in the form of linear combination of potential targets with an over complete dictionary. In previous studies, the antennas of ground penetrating radars (GPRs) are assumed to be isotropic when generating the dictionary entries, however, the beam width of the antennas are finite in practice. Thus, modeling errors would be introduced when the interested imaging region covers a wide range in the cross-range direction. To reduce this modeling error, a modified dictionary which takes the antenna patterns into account is proposed in this work. Simulated results show that the image reconstructed by utilizing the modified dictionary achieves a superior performance compared to the classic one.