Meta-surface lens and deep learning process toward practical LWIR imaging devices

Byungkyu Lee,Jangwoo You,Soonkeun Chang,Duhyun Lee,Seyoon Kim,Yongseop Yoon,Wontaek Seo

DOI: https://doi.org/10.1117/12.2674111

2023-10-10

Abstract:We report the useful approach toward the practical imaging solution for long-wavelength infrared (LWIR). In order to make use of the LWIR sensors conveniently in everyday life, the imaging module needs to be slim and small so that it can be mounted inside end-user devices without big difficulties. At the same time, the image qualities should have a sufficient level to guarantee that people can easily identify object shapes and recognize temperature differences when they see the resultant images. In this paper, we focus on those two crucial points for the practical LWIR imaging device. First, to make the compact optical system, we adopted a thin meta-surface lens of a focal length 2 mm which is showing the effective total top length (TTL) less than 3 mm. Second, to enhance the image sharpness degraded relatively due to the lens, the deep learning method of the U-net model is introduced. The patterns of the USAF resolution target chart indicate the increase of modulation value by 3~8 times after applying the learning process. We believe that our work helps to expand the pragmatic application area of the LWIR imaging sensors in the near future.

Engineering,Computer Science,Physics