Wang Ze, Chen Hang, Li Jianan, Xu Tingfa, Zhao Zejia, Duan Zhengyang, Gao Sheng, Lin Xing
School of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, China.
Department of Electronic Engineering, Tsinghua University, Beijing 100084, China.
Nanophotonics. 2024 Jul 2;13(20):3883-3893. doi: 10.1515/nanoph-2024-0233. eCollection 2024 Aug.
Spectral reconstruction, critical for understanding sample composition, is extensively applied in fields like remote sensing, geology, and medical imaging. However, existing spectral reconstruction methods require bulky equipment or complex electronic reconstruction algorithms, which limit the system's performance and applications. This paper presents a novel flexible all-optical opto-intelligence spectrometer, termed OIS, using a diffractive neural network for high-precision spectral reconstruction, featuring low energy consumption and light-speed processing. Simulation experiments indicate that the OIS is able to achieve high-precision spectral reconstruction under spatially coherent and incoherent light sources without relying on any complex electronic algorithms, and integration with a simplified electrical calibration module can further improve the performance of OIS. To demonstrate the robustness of OIS, spectral reconstruction was also successfully conducted on real-world datasets. Our work provides a valuable reference for using diffractive neural networks in spectral interaction and perception, contributing to ongoing developments in photonic computing and machine learning.
光谱重建对于理解样本成分至关重要，在遥感、地质和医学成像等领域有着广泛应用。然而，现有的光谱重建方法需要庞大的设备或复杂的电子重建算法，这限制了系统的性能和应用。本文提出了一种新颖的灵活全光光智能光谱仪，称为OIS，它使用衍射神经网络进行高精度光谱重建，具有低能耗和光速处理的特点。仿真实验表明，OIS能够在空间相干和非相干光源下实现高精度光谱重建，无需依赖任何复杂的电子算法，并且与简化的电校准模块集成可以进一步提高OIS的性能。为了证明OIS的鲁棒性，还在真实世界数据集上成功进行了光谱重建。我们的工作为在光谱交互和感知中使用衍射神经网络提供了有价值的参考，有助于光子计算和机器学习的持续发展。
