Deep-learning-driven end-to-end metalens imaging

Joonhyuk Seo,Jaegang Jo,Joohoon Kim,Joonho Kang,Chanik Kang,Seongwon Moon,Eunji Lee,Jehyeong Hong,Junsuk Rho,Haejun Chung

2024-05-10

Abstract:Recent advances in metasurface lenses (metalenses) have shown great potential for opening a new era in compact imaging, photography, light detection and ranging (LiDAR), and virtual reality/augmented reality (VR/AR) applications. However, the fundamental trade-off between broadband focusing efficiency and operating bandwidth limits the performance of broadband metalenses, resulting in chromatic aberration, angular aberration, and a relatively low efficiency. In this study, a deep-learning-based image restoration framework is proposed to overcome these limitations and realize end-to-end metalens imaging, thereby achieving aberration-free full-color imaging for mass-produced metalenses with 10-mm diameter. Neural-network-assisted metalens imaging achieved a high resolution comparable to that of the ground truth image.

Optics,Image and Video Processing

What problem does this paper attempt to address?

The problems that this paper attempts to solve are: how to overcome the problems such as chromatic aberration, angular aberration and relatively low efficiency faced by metalenses in broadband imaging, and achieve high - quality, aberration - free full - color imaging. Specifically, the researchers proposed an end - to - end image restoration framework based on deep learning, aiming to improve the image quality obtained by metalenses through this framework, making it close to or reach the quality level of the real image. This framework is especially suitable for 10 - millimeter - diameter metalenses in large - scale production, and can achieve high - resolution full - color imaging while maintaining a compact design.

The paper mentions that although metalenses have attracted much attention because of their ultrathin structure and the ability to overcome the inherent limitations of traditional lenses (such as chromatic aberration, spherical aberration, large volume and high manufacturing cost), in the pursuit of broadband imaging performance, there are still multi - aspect trade - off problems among focusing efficiency, lens diameter and spectral bandwidth. These problems have led to the chromatic aberration or low focusing efficiency under large bandwidth of the currently reported broadband metalenses, limiting the possibility of their commercial applications. Therefore, through the introduction of deep - learning technology and in combination with a customized image restoration framework, this research proposes a comprehensive solution, which not only effectively solves the above - mentioned aberration problems, but also takes advantage of the large - scale production of metalenses, opening up a new path for the realization of high - quality, aberration - free compact imaging systems.