Abstract
A deep-learning-based phase retrieval method is proposed for computational ghost imaging (CGI). In the method, a defocused intensity distribution of an object is obtained by the CGI. An object’s phase distribution is retrieved from the intensity distribution by a trained deep neural network. The advantage of the method is that the phase retrieval can be achieved with a low signal to noise ratio. The effects of noise level and defocus distance are investigated.
© 2018 Optical Society of Japan
PDF ArticleMore Like This
Alexandre Goy, Kwabena Arthur, Shuai Li, and George Barbastathis
FTh3E.5 Frontiers in Optics (FiO) 2018
Jianying Hao, Ruixian Chen, Xiao Lin, Tsutomu Shimura, and Xiaodi Tan
IPDP_01 International Symposium on Imaging, Sensing, and Optical Memory (ODS) 2022
Xiaofeng Wu, Sibi Chakravarthy Shanmuagvel, and Yunhui Zhu
CM2A.3 Computational Optical Sensing and Imaging (COSI) 2022