Direct field-to-pattern monolithic design of holographic metasurface via residual encoder-decoder convolutional neural network
基于残差编码器-解码器卷积神经网络的全息元表面直接场到图案单片设计
残差エンコーダ−デコーダ畳み込みニューラルネットワークに基づくホログラフィック要素表面直接場対パターンモノリシック設計
잔차 인코더-디코더 볼륨 신경망 기반 홀로그램 표면 직접 필드 도안 단편 설계
Diseño de un solo chip de campo directo a patrón en la superficie del elemento holográfico basado en la red neuronal de convolución del codificador y decodificador residual
Conception monolithique de champ direct à motif de méta - surface holographique basée sur un réseau neuronal convolutif codeur - décodeur résiduel
На основе остаточного кодера - декодера свернуть голографическую поверхность нейронной сети прямо в рисунок монолитного дизайна
Ruichao Zhu 朱瑞超 ¹, Jiafu Wang 王甲富 ¹, Tianshuo Qiu 邱天硕 ¹, Dingkang Yang 杨鼎康 ², Bo Feng 封波 ¹, Zuntian Chu 楚遵天 ¹, Tonghao Liu 刘同豪 ¹, Yajuan Han 韩亚娟 ¹, Hongya Chen 陈红雅 ¹, Shaobo Qu 屈绍波 ¹
¹ Shaanxi Key Laboratory of Artificially-Structured Functional Materials and Devices, Air Force Engineering University, Xi'an 710051, China
中国 西安 中国人民解放军空军工程大学 陕西省人工结构功能材料与器件重点实验室
² The Academy for Engineering & Technology, Fudan University, Shanghai 200433, China
中国 上海 复旦大学工程与应用技术研究院
Complex-amplitude holographic metasurfaces (CAHMs) with the flexibility in modulating phase and amplitude profiles have been used to manipulate the propagation of wavefront with an unprecedented level, leading to higher image-reconstruction quality compared with their natural counterparts. However, prevailing design methods of CAHMs are based on Huygens-Fresnel theory, meta-atom optimization, numerical simulation and experimental verification, which results in a consumption of computing resources.
Here, we applied residual encoder-decoder convolutional neural network to directly map the electric field distributions and input images for monolithic metasurface design. A pretrained network is firstly trained by the electric field distributions calculated by diffraction theory, which is subsequently migrated as transfer learning framework to map the simulated electric field distributions and input images. The training results show that the normalized mean pixel error is about 3% on dataset.
As verification, the metasurface prototypes are fabricated, simulated and measured. The reconstructed electric field of reverse-engineered metasurface exhibits high similarity to the target electric field, which demonstrates the effectiveness of our design. Encouragingly, this work provides a monolithic field-to-pattern design method for CAHMs, which paves a new route for the direct reconstruction of metasurfaces.
