Search
Menu
Oxford Instruments WITec GmbH - Raman Microscope LB 12/25
Photonics HandbookResearch & Technology

Multi-Layer Diffractive Optical Processors Enable Unidirectional Visible Imaging

Facebook X LinkedIn Email
LOS ANGELES, Aug. 20, 2025 — Used in optical computing and computational imaging, diffractive optical processors and metasurfaces are flourishing within the computing industry. The demonstrations of these materials are constrained to 2D implementations and longer wavelengths due to the fabrication challenges of nanoscale features in 3D diffractive architectures.

Researchers at the UCLA Samueli School of Engineering and the Optical Systems Division at Broadcom Inc. have developed a broadband, polarization-insensitive unidirectional imager that operates in the visible spectrum, capable of high-efficiency image transmission in one direction while effectively suppressing image formation in the reverse direction. This device incorporates diffractive structures fabricated through wafer-scale lithography on high-purity fused silica, offering high optical transparency, thermal stability, and ultra-low loss.

The researchers demonstrated that their nano-fabrication approach, along with deep learning-based inverse design, enables visible image formation in only one direction, transmitting images from the input field of view to the output field of view — while blocking and distorting image formation in the reverse direction. According to the researchers, the work represents the first demonstration of broadband unidirectional imaging in the visible spectrum, achieved with nanoscale, polarization-insensitive diffractive features that were optimized using deep learning.

The developed fabrication process leverages methods compatible with semiconductor manufacturing, opening the potential to optoelectronic components. This work marks significant advancement towards future advances in computational imaging, optical sensing, and optical information processing, with potential applications in compact multispectral imagers and optical privacy protection.

This research was published in Light Science & Applications (www.doi.org/10.1038/s41377-025-01971-2).
Stanford Research Systems - Precision DC Voltage 3-25 300x250

Published: August 2025
Glossary
deep learning
Deep learning is a subset of machine learning that involves the use of artificial neural networks to model and solve complex problems. The term "deep" in deep learning refers to the use of deep neural networks, which are neural networks with multiple layers (deep architectures). These networks, often called deep neural networks or deep neural architectures, have the ability to automatically learn hierarchical representations of data. Key concepts and components of deep learning include: ...
computational imaging
Computational imaging refers to the use of computational techniques, algorithms, and hardware to enhance or enable imaging capabilities beyond what traditional optical systems can achieve. It involves the integration of digital processing with imaging systems to improve image quality, extract additional information from captured data, or enable novel imaging functionalities. Principles: Computational imaging combines optics, digital signal processing, and algorithms to manipulate and...
Research & TechnologymanufacturingsemiconductorsMaterialscomputingdiffractiveoptical processorunidirectionalImagingUCLASamueli School of EngineeringBroadcomOptical Systems Divisiondeep learningfabricationnanofabricationcomputational imagingsensingmetasurfaceAmericas

We use cookies to improve user experience and analyze our website traffic as stated in our Privacy Policy. By using this website, you agree to the use of cookies unless you have disabled them.