Opto-intelligence spectrometer using diffractive neural networks

Abstract

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.