Abstract

With the development of computational imaging, the integration of optical system design and digital algorithms has made more imaging tasks easier to perform. Wavefront coding (WFC) is a typical computational imaging technique that is used to address the constraints of optical aperture and depth of field. In this paper, we demonstrated a low-cost and simple optical system based on WFC and deep learning. We constructed an optimized encoding method for the phase plate under the framework of deep learning, which reduces the requirement for aberration correction in the full field of view. Optical coding was achieved with just a double-bonded lens and a simple cubic phase mask, and digital decoding used the deep residual UNet$++$ network framework. The final image obtained has good resolution, whereas the depth of field of the system expanded by a factor of 13, which is of great significance for the high-precision inspection and attaching of small parts of machine vision.

© 2023 Optica Publishing Group

Extended depth-of-field infrared imaging with deeply learned wavefront coding

Yidan Li, Junhua Wang, Xintong Zhang, Kai Hu, Lu Ye, Minge Gao, Yuxiang Cao, and Min Xu
Opt. Express 30(22) 40018-40031 (2022)

Wavefront coding image reconstruction via physical prior and frequency attention

Qinghan Zhang, Meng Bao, Liujie Sun, Yourong Liu, and Jihong Zheng
Opt. Express 31(20) 32875-32886 (2023)

Imaging and image restoration of an on-axis three-mirror Cassegrain system with wavefront coding technology

Xiaohu Guo, Liquan Dong, Yuejin Zhao, Wei Jia, Lingqin Kong, Yijian Wu, and Bing Li
Appl. Opt. 54(10) 2798-2805 (2015)

Experimental study of an off-axis three mirror anastigmatic system with wavefront coding technology

Feng Yan and Xiaoping Tao
Appl. Opt. 51(11) 1749-1756 (2012)

4-K-resolution minimalist optical system design based on deep learning

Dexiao Meng, Yan Zhou, and Jian Bai
Appl. Opt. 63(4) 917-926 (2024)