Phase Reconfigurability-Enhanced Metasurface Image Compression with an Integrated Photonic Encoder

Nanophotonic encoders compress raw image data with low latency and energy consumption. On the integrated photonic platform, passive disordered subwavelength structures provide random projection as an analogue encoder, which is a programming energy-free approach to image compression. Here, we show that introducing reconfigurability can significantly improve the fidelity and further the speed and energy advantages of the analogou photonic processor. We built a multichannel phase-reconfigurable waveguiding metasystem on a single chip, with a phase array embedded between diffractive layers. With the optimized phase, the metasystem can compress a 512 × 512-pixel image, achieving a peak signal-to-noise ratio (PSNR) of 36 dB and a structural similarity of 0.97, compared to a PSNR of 6.24 dB and a similarity of 0.046 with a passive structure only.