Abstract: Considering the sparse structure of actual target space in GPR application, a novel 3D imaging method based on random aperture compressive sensing is proposed in this paper, which capable of reconstructing the target space from a few compressive sensing data obtained by random aperture measurements, and the imaging performance versus noise level and the effects of different measurement matrices are analyzed. The computer simulation results indicate that the proposed algorithm allow much fewer data, much shorter measurement time. And due to it is fully utilization of the sparse structure of interested target space, the method show much more robust and sparse image than time-domain standard back-projection method.