WANG Zesheng;MENG Feiyang;YANG Mingrui;LIN Zhihao;LIU Junxian;SUN Wenbo;

• 1:Department of Physics
• 2:Tsinghua University
• 3:Department of Mechanical Engineering
• 4:Tsinghua University

Abstract：

Convolutional neural networks are widely used in tasks such as image recognition due to their excellent performance in two-dimensional data processing.To address the limited AI computing power under the traditional digital integrated circuit computing architecture,this study designs an optical convolution system based on a microlens array.Utilizing the ultra-high-speed and high-parallelism characteristics of optical computing,it achieves convolution computation based on the principles of geometrical optics.The system loads image information and convolution kernels using a spatial light modulator,segments the light field through a microlens array,modulates the segmented sub-regions in parallel using the convolution kernel,and finally converges through the imaging optical path to achieve summation,thuscompletingtheconvolutioncomputationinopticalspace.Thisstudypresentstheimplementationprincipleoftheopticalconvolutionsystembasedonamicrolensarray,designs the experimental optical path,and proposes implementation methods for key technologies such as system calibration,convolution kernel implementation,and image post-processing.Experimental results show that the system possesses basic imaging performance and the ability to implement complex feature operators such as Gaussian operator,Gradient operator,and Laplacian operator.Limited by device accuracy,the system still has errors caused by aberrations and diffraction.It can provide verification for the feasibility of optical convolution schemes based on geometrical optical architecture in image preprocessing tasks.

Key Words： optical information processing;convolutional neural network;microlens array;image processing

Abstract:

Keywords:

Foundation: 教育部拔尖计划2.0研究课题（20211007）;教育部产学合作协同育人项目（231001318115817）;; 清华大学实验室创新基金（202504303）

Authors: WANG Zesheng;MENG Feiyang;YANG Mingrui;LIN Zhihao;LIU Junxian;SUN Wenbo;

DOI: 10.27024/j.wlygc.2026.06.24.02

References：

• [1] KRIZHEVSKY A,SUTSKEVER I,HINTON G E.Imagenet classification with deep convolutional neural networks[J].Advances in neural information processing systems,2012:25.
• [2] CHEN Y H,KRISHNA T,EMER J S,et al.Eyeriss:An energy-efficient reconfigurable accelerator for deep convolutional neural networks[J].IEEE journal of solid-state circuits,2016,52(1):127-138.
• [3] 周志华.机器学习[M].北京：清华大学出版社，2016.ZHOU Z H.Machine learning[M].Beijing:Tsinghua University Press,2016.(in Chinese)
• [4] LIN X,RIVENSON Y,YARDIMCI N T,et al.All-optical machine learning using diffractive deep neural networks[J].Science,2018,361(6406):1004-1008.
• [5] 费宇航，隋修宝，王庆宝，等.微透镜阵列光学实现卷积运算[J].红外与激光工程，2022,51(2):425-430.FEI Y H,SUI X B,WANG Q B,et al.Optically realize convolution operation of microlens array[J].Infrared and Laser Engineering,2022,51(2):425-430.(in Chinese)
• [6] 费宇航.低功耗光学图像分类神经网络研究[D].南京理工大学，2023.FEI Y H.Research on low-power optical image classification neural network[D].Nanjing University of Science and Technology,2023.(in Chinese)
• [7] 周浩军，周海龙，董建绩.光学卷积计算的进展与挑战(特邀)[J].光学学报，2024,44(15):374-389.ZHOU H J,ZHOU H L,DONG J J.Advances and challenges of optical convolution computation (invited)[J].Acta Optica Sinica,2024,44(15):374-389.(in Chinese)
• [8] DEV K,ASUNDI A.Mueller-Stokes polarimetric characterization of transmissive liquid crystal spatial light modulator[J].Optics and Lasers in Engineering,2012,50(4):599-607.
• [9] MARR D,HILDRETH E.Theory of edge detection[J].Proceedings of the Royal Society of London.Series B.Biological Sciences,1980,207(1167):187-217.
• [10] KANOPOULOS N,VASANTHAVADA N,BAKER R L.Design of an image edge detection filter using the Sobel operator[J].IEEE Journal of solid-state circuits,1988,23(2):358-367.