Terahertz meta-holograms reconstruction based on compressed sensing

The introduction of single-pixel imaging and compressed sensing (CS) techniques into terahertz (THz) imaging has sped up image acquisition and avoided raster scanning. At present, most single-pixel terahertz imaging developments are based on simple metallic samples, researchers rarely study the reconstruction of complex structural samples with large attenuation in the terahertz domain, such as metasurface holographic images. Here, we present an implementation of the single-pixel compressed sensing approach into THz metasurface holography reconstruction. By laser projecting a set of binary patterns on a 500-m thick silicon wafer using a digital micromirror device (DMD), THz wavefront of the holographic metasurface is spatially encoded. Single-element detector is used to measure electric field amplitude of the transmitted THz radiation for each pattern, and then the hologram is reconstructed by Total variation Augmented Lagrangian and Alternating Direction Algorithm (TVAL3). Besides, the reconstruction effects are also analyzed by reducing measurement number, it can maintain more than 95% of the image information under 20% compression. The demonstrated combination of terahertz holography and single-pixel compressed sensing imaging provides new possibilities for metasurface imaging, verifies the stability of terahertz single-pixel imaging, and the scheme may lead to advances in fast terahertz imaging.