A novel concept for suppressing the problem of multipath ghosts in Ground Penetrating Radar (GPR) and Through-Wall Imaging (TWI) is presented. Ghosts (i.e., false targets) mainly arise from the use of the Born or single-scattering approximations that lead to linearized imaging algorithms; however, these approximations neglect the effect of multiple scattering (or multipath) between the electromagnetic wavefield and the object under investigation. In contrast to existing methods of suppressing multipath ghosts, the proposed method models for the first time the reflectivity of the probed objects as a probability function up to a normalized factor and introduces the concept of random subaperture by randomly picking up measurement locations from the entire aperture. Thus, the final radar image is a joint probability distribution that corresponds to radar images derived from multiple random subapertures. Finally, numerical experiments are used to demonstrate the performance of the proposed methodology in GPR and TWI imaging.