End-to-end Optimization of Metalens for Wide-angle and Broadband Imaging

Yeongmyeong Park, Youngjin Kim,Yoonchan Jeong, Changhyun Kim, Gun-Yeal Lee, Hyeongu Choi, Taewon Choi, Byoungho Lee

End-to-end optimization of metalens and artificial intelligence-driven image restoration algorithms has recently emerged as a powerful tool for realizing ultra-compact imaging systems. However, the limited imaging quality of existing approaches remains challenging in meeting the demand for commercial devices due to the severe aberrations exhibited by metalens. These results in highly blurred sensor images, creating substantial challenges for accurate image restoration. In this work, a novel meta-imager is introduced that overcomes this challenge by employing an aperture-stop-integrated metalens and co-designing it with a computational image restoration network using a fully differentiable optimization framework. The proposed imager physically consists of a single metalens and an aperture stop located on the opposite side of the 1 mm-thick glass substrate. This configuration effectively alleviates off-axis aberrations such as coma and astigmatism, facilitating the image restoration process of the deep neural networks. The experimental results present that this scheme features 70° field-of-view, for full-color imaging across the entire visible spectrum. It is believed that this work represents a significant advancement in creating ultra-compact cameras using nanophotonics.