Non-line-of-sight Imaging with Radio Signals

Non-line-of-sight (NLOS) imaging reconstructs the shape and albedo of objects outside the field of view of an imaging system. Based on this technique, a wide range of applications such as autonomous vehicle navigation, robotic and machine vision, as well as medical imaging will possess unprecedented capabilities. Recent NLOS imaging approaches employ ultra-fast pulsed laser and time-resolved photon detector to collect light transients, which are however time-consuming due to the point-by-point grid scan on the wall. Moreover, the maximum imaging distance is only a few meters because of the limitation of the photon counter. In this paper, we introduce Radio Frequency (RF) NLOS imaging to reconstruct hidden objects with radio signals. By analyzing the image formation model, we transform the reconstruction problem into an optimization problem that can be solved with additional constraints. The simulation results validate that we can reconstruct letter-shaped objects using a 12×12 multipleinput multiple-output (MIMO) array and a stepped frequency signal with a bandwidth of 4GHz., Compared with the optical NLOS imaging, the proposed method can capture the hidden 3D geometry at longer ranges with shorter acquisition times.