Reconfigurable Metasurface Hologram of Dynamic Distance via Deep Learning

Abstract

Reconfigurable metasurfaces have been regarded as an emerging subfield of metasurfaces that can manipulate electromagnetic wave information in a smart manner. They stimulate a gradual transition in metasurface holography from passive to active elements. To date, intelligent dynamic holographic imaging schemes typically rely on iterative or data-driven methods to obtain holograms at a fixed imaging distance, which significantly hinders the development of intelligent dynamic holographic imaging in practical scenarios involving high demands for dynamic imaging distances. Herein, a computer-generated hologram algorithm with a dynamic imaging distance and a reconfigurable metasurface are proposed, which is referred to as a generator and physical diffractive network. Simulation results of time–distance division for three-dimensional imaging are provided to demonstrate the reliability and high efficiency of the proposed algorithm.