Machine-learning reprogrammable metasurface imager

• Published in: Nature Communications Vol. 10; no. 1; pp. 1082 - 8
• Main Authors: Li, Lianlin, Ruan, Hengxin, Liu, Che, Li, Ying, Shuang, Ya, Alù, Andrea, Qiu, Cheng-Wei, Cui, Tie Jun
• Format: Journal Article
• Language: English
• Published: London Springer Science and Business Media LLC 06.03.2019; Nature Publishing Group UK; Nature Publishing Group; Nature Portfolio
• Subjects: 639/166/987; 639/624/399/1015; 639/766/930/2735; Algorithms; Data acquisition; Data processing; Data transfer (computers); Digits; Humanities and Social Sciences; Image coding; Image reconstruction; Learning algorithms; Machine learning; Metasurfaces; multidisciplinary; Object recognition; Post-production processing; Q; Real time; Science; Science (multidisciplinary)
• ISSN: 2041-1723; 2041-1723
• DOI: 10.1038/s41467-019-09103-2

Conventional microwave imagers usually require either time-consuming data acquisition, or complicated reconstruction algorithms for data post-processing, making them largely ineffective for complex in-situ sensing and monitoring. Here, we experimentally report a real-time digital-metasurface imager that can be trained in-situ to generate the radiation patterns required by machine-learning optimized measurement modes. This imager is electronically reprogrammed in real time to access the optimized solution for an entire data set, realizing storage and transfer of full-resolution raw data in dynamically varying scenes. High-accuracy image coding and recognition are demonstrated in situ for various image sets, including hand-written digits and through-wall body gestures, using a single physical hardware imager, reprogrammed in real time. Our electronically controlled metasurface imager opens new venues for intelligent surveillance, fast data acquisition and processing, imaging at various frequencies, and beyond. Conventional imagers require time-consuming data acquisition, or complicated reconstruction algorithms for data post-processing. Here, the authors demonstrate a real-time digital-metasurface imager that can be trained in-situ to show high accuracy image coding and recognition for various image sets.