In the age of data-driven technologies, optical computing—which processes information using light instead of electricity—has emerged as a promising frontier. Unlike traditional electronic processors, optical systems can perform computations at the speed of light with extremely low energy consumption. This opens up exciting opportunities in fields such as artificial intelligence (AI), autonomous vision, and real-time medical imaging.
A recent collaborative study between Professor LI Wei's group at the Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences and Professor Andrea Alù's group at the City University of New York represents a significant leap in optical computing. Published in Nature Communications, the work introduces a nonlocal flat optical device capable of real-time, size-selective image processing and optical denoising, marking a notable first in this area of research. Historically, optical systems for image processing have relied on bulky lenses or complex Fourier optics setups. While metasurfaces and nanophotonic devices have been explored as miniaturized alternatives, these designs have often lacked multifunctionality. Many were not cost-effective, scalable, or able to dynamically handle noisy input data—particularly where background interference or size-based object selection is involved. What sets this new device apart is its ability to perform spatial band-pass filtering in momentum space using a remarkably simple structure: a three-layer metal-dielectric-metal cavity. This compact configuration not only enhances edge detection down to ~0.9 μm but, for the first time, integrates image denoising capabilities through optical transfer function (OTF) engineering—a technique rarely explored for noise reduction in all-optical systems.
Unlike previous devices that lacked selectivity, the nonlocal flat optics developed here can distinguish features based on size—effectively acting as a size-sensitive filter. This allows the system to reject unwanted background noise while preserving critical image details. Such functionality is crucial for applications ranging from biomedical imaging to autonomous navigation, where clarity and precision are non-negotiable. Furthermore, the design is scalable and cost-effective, making it a strong candidate for commercial and industrial use. It’s simple three-layer structure lends itself to mass manufacturing, paving the way for integration into compact imaging and sensing systems.
This work not only pushes the boundaries of flat optics but also opens new pathways for optical AI hardware, where light itself performs the computations traditionally handled by digital processors. Future research could explore tunable or reconfigurable versions of the device, adaptive filtering for dynamic environments, or integration with AI systems to enable smarter, faster, and more energy-efficient decision-making—all powered by light. As the demand for real-time, low-power computation grows, innovations like this represent a critical step toward the future of optical computing and intelligent vision systems.
中文
