All-optical computing provides a promising route to real-time, energy-efficient image processing by performing computations directly with light. However, conventional optical processors are often bulky with limited functionality, making them difficult to integrate into compact imaging systems.
In a new paper published in Light: Science & Applications , Linzhi Yu and his supervisor Humeyra Caglayan put forward a powerful yet compact solution: “meta-operators” that perform all-optical image processing and complex holography using a single passive layer. The approach integrates double-phase encoding with polarization multiplexing to achieve full complex-field modulation at visible wavelengths.
This platform enables real-time computational tasks such as edge detection, object recognition, and 3D hologram reconstruction, without digital post-processing. The researchers demonstrate first- and second-order differentiation, cross-correlation, and volumetric holography using engineered nanopillar arrays. The metasurface operators also maintain robust performance under misalignment and multi-wavelength illumination.
This work opens new opportunities for miniaturized analog optical computing, with immediate relevance for imaging, sensing, and next-generation holographic display technologies. It represents a significant step toward bridging computational optics with practical nanophotonic hardware.
Light Science & Applications
Double-phase metasurface operators for all-optical image processing