An Active Metasurface Single-Pixel Lensless Imaging Device

We present a computational model that evaluates the performance of active metasurfaces as platforms for lensless imaging and find that an active metasurface with a 0.3 mm aperture size could capture images at &gt;2 pixel/degree resolution across a wide field of view (FOV) of 120<i>°</i>. Our model optimizes the metasurface array configuration to obtain angular Hadamard and Fourier bases, and the results address the performance of an experimentally realized indium tin oxide (ITO)-based metasurface operating at telecom wavelengths.<br/><br/>Lens-coupled detector arrays are the basis for most of today’s imaging systems but are not well suited to realized low-weight or small-footprint systems due to the weight and focal lengths of lenses. Lensless imaging—in which a modulator is coupled to a detector and modifies the detected light to enable computational image retrieval—offers an attractive alternative to lens-coupled detectors in applications requiring low SWaP (size, weight and power) devices or wide FOVs [1].<br/><br/>We investigate active metasurface elements to modulate light in a single-pixel imaging configuration. In contrast to passive metasurfaces that have set optical properties, the phases and amplitudes of active metasurface elements can be dynamically controlled via external stimuli which modulate their dielectric properties. In this study, we use the experimentally demonstrated performance of a field-effect tunable ITO-based metasurface at an operating wavelength of 1510 nm, with a phase tunability of 271<i>°</i>, covarying reflectance and phase, and modulation speeds exceeding 10 MHz [2][3]. The model and analysis, however, can be applied to various active metasurface platforms.<br/><br/>To determine the attainable performance of active metasurfaces in imaging, we use array-level inverse design, which was previously used to enhance beam steering performance [4]. We simulate the 1D far-field intensity of the metasurface using an array factor calculation and control the acceptance of light by optimizing the voltages applied at each element. Using a simulated aperture size of 0.3 mm (768 elements), we demonstrate a peak resolution of 4 pixel/degree and of 2 pixel/degree across a 120<i>° </i>FOV. We also show the suitability of active metasurfaces for compressed sensing, in which sparse images are recovered in fewer measurements than there are pixels. For this, we optimize the metasurface array configuration to obtain angular Hadamard and Fourier bases. In preliminary tests, our optimized Hadamard basis retrieves images with good agreement to the ground truth, with a power signal-to-noise ratio of 40 dB across 512 pixels. Additionally, we demonstrate the convolution of image processing filters across the FOV, which is desirable in classification applications. Our ability to produce grayscale filters distinguishes our platform from digital micromirror devices, which are often used in single-pixel imaging and support modulation speeds of ~20 kHz only for binary bases [5]. As resolution and FOV improve with metasurface size, we note that our current results are not an upper bound on performance.<br/><br/>As a next step, we will explore algorithms for 2D imaging and increase the realism of our model by modeling amplitude/phase errors, voltage discretization, and component reaction times. We will also report on an analysis of which imaging bases are best suited to active metasurfaces, and on the use of machine learning to correct for imperfections in basis construction. Finally, we will discuss the potential of active metasurfaces in multi-pixel lensless imaging applications.<br/><br/>[1] V. Boominathan<i> et al.</i>, <i>Optica</i>, vol. 9, no. 1, p. 1, 2022.<br/>[2] Y.-W. Huang <i>et al.</i>, <i>Nano Lett.</i>, vol. 16, no. 9, pp. 5319–5325, 2016.<br/>[3] G. K. Shirmanesh <i>et al.</i>, <i>ACS Nano</i>, vol. 14, no. 6, pp. 6912–6920, 2020.<br/>[4] P. Thureja <i>et al.</i>, <i>ACS Nano</i>, vol. 14, no. 11, pp. 15042–15055, 2020.<br/>[5] G. M. Gibson <i>et al.</i>, <i>Opt. </i><i>Express</i>, vol. 28, no. 19, p. 28190, 2020.