Based on frequency selective surface (FSS) structure, a millimeter wave compressed sensing imaging method is proposed, which provides the research approach for hardware realization of compressed sensing imaging. By loading switching diodes into FSS and making them on/off randomly, a new random-switching millimeter wave imaging mask is simulated and designed. The mask designed is put on a millimeter wave antenna. The corresponding random measurement matrix can be constructed and enough effective measurement times can be obtained. Combined with compressed sensing theory, the imaging simulation experiments are completed by using reconstruction algorithm. Results confirm feasibility of the method proposed, and reconstruction of the original images can be achieved in the case of lower sampling rate.