Abstract
We demonstrate that deep neural networks can be trained on object-intensity pairs to efficiently and accurately produce object estimates from unseen raw intensity images at test time, thus solving the inverse problem.
© 2017 Optical Society of America
PDF ArticleMore Like This
Jiapeng Liu, Muralidhar M. Balaji, Christopher A. Metzler, M. Salman Asif, and Prasanna Rangarajan
CTh5A.2 Computational Optical Sensing and Imaging (COSI) 2021
Gili Dardikman and Natan T. Shaked
CW3B.5 Computational Optical Sensing and Imaging (COSI) 2018
John Peurifoy, Yichen Shen, Yi Yang, Li Jing, Fidel Cano-Renteria, John Joannopoulos, Max Tegmark, and Marin Soljačić
FTh4A.4 Frontiers in Optics (FiO) 2017