收藏切换
Image−driven metasurface inverse design via multi−model collaborative artificial intelligence
收藏切换
PDF
Liming SI1, 2, 3, 4, 5, Rong NIU1, 2, 3, Chenyang DANG1, 2, 3, Zhaorui WANG1, 2, 3, Tianyu MA1, 2, 3, Yan LI5, Weiren ZHU6, Houjun SUN1, 2, 3
Science & Technology Review | 2026, 44(9) : 88 - 97
Less
收藏切换
Science & Technology Review | 2026, 44(9): 88-97
Exclusive
Image−driven metasurface inverse design via multi−model collaborative artificial intelligence
Full
Liming SI1, 2, 3, 4, 5, Rong NIU1, 2, 3, Chenyang DANG1, 2, 3, Zhaorui WANG1, 2, 3, Tianyu MA1, 2, 3, Yan LI5, Weiren ZHU6, Houjun SUN1, 2, 3
Affiliations
  • 1School of Integrated Circuits and Electronics, Beijing Institute of Technology, Beijing 100081, China
  • 2State Key Laboratory of Environment Characteristics and Effects for Near−space, Beijing 100081, China
  • 3Tangshan Research Institute, Beijing Institute of Technology, Tangshan 063000, China
  • 4State Key Laboratory of Millimeter Waves, Nanjing 210096, China
  • 5Faculty of Engineering, Shenzhen MSU−BIT University, Shenzhen 518172, China
  • 6School of Integrated Circuits, Shanghai Jiao Tong University, Shanghai 200240, China
Published: 2026-05-13 doi: 10.3981/j.issn.1000-7857.2026.01.00116
Outline
收藏切换

With the rapid development of metasurfaces for precise electromagnetic wave manipulation, the efficient and automated design of complex functional devices has become a critical research challenge. This paper proposes a multi−model collaborative framework for metasurface inverse design driven by visual images, aiming to automatically generate electromagnetic metasurface array structures directly from visual image inputs. The method employs a synergistic multi−model strategy to establish an end−to−end mapping from image semantic features to physical metasurface structural parameters. Specifically, a conditional generative adversarial network (Pix2Pix) is first employed to transform visual images into target holographic textures, followed by a U−shaped convolutional neural network (U−Net) for high−fidelity phase distribution prediction. Subsequently, a variational autoencoder (VAE) is introduced to map the desired phase profiles to manufacturable unit−cell geometries, effectively bridging the long−standing gap between phase synthesis and structural modeling in conventional design workflows. Numerical simulations demonstrate that the designed holographic metasurfaces can accurately reconstruct target holographic patterns at designated operating frequencies, achieving significantly improved reconstruction fidelity and design stability compared with traditional single−network approaches. This work provides an efficient and automated new paradigm for image−driven intelligent metasurface design and offers a promising solution for the rapid development of complex electromagnetic functional devices.

artificial intelligence  /  multi−model collaborative framework  /  metasurfaces  /  inverse design  /  visual images  /  neural networks
Liming SI, Rong NIU, Chenyang DANG, Zhaorui WANG, Tianyu MA, Yan LI, Weiren ZHU, Houjun SUN. Image−driven metasurface inverse design via multi−model collaborative artificial intelligence[J]. Science & Technology Review, 2026 , 44 (9) : 88 -97 . DOI: 10.3981/j.issn.1000-7857.2026.01.00116
Year 2026 volume 44 Issue 9
PDF
195
105
Cite this Article
BibTeX
Article Info
doi: 10.3981/j.issn.1000-7857.2026.01.00116
  • Receive Date:2026-01-21
  • Online Date:2026-05-27
  • Published:2026-05-13
Article Data
Affiliations
History
  • Received:2026-01-21
  • Revised:2026-03-30
Funding
Affiliations
    1School of Integrated Circuits and Electronics, Beijing Institute of Technology, Beijing 100081, China
    2State Key Laboratory of Environment Characteristics and Effects for Near−space, Beijing 100081, China
    3Tangshan Research Institute, Beijing Institute of Technology, Tangshan 063000, China
    4State Key Laboratory of Millimeter Waves, Nanjing 210096, China
    5Faculty of Engineering, Shenzhen MSU−BIT University, Shenzhen 518172, China
    6School of Integrated Circuits, Shanghai Jiao Tong University, Shanghai 200240, China
References
Share
https://castjournals.cast.org.cn/joweb/kjdb/EN/10.3981/j.issn.1000-7857.2026.01.00116
Share to
QR

Scan QR to access full text

Cite this article
BibTeX
Citations
表12种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏
  • BibTeX
  • EndNote
  • RefWorks
  • TxT