|
|
|
|
| LEADER |
01635 am a22002293u 4500 |
| 001 |
136711.2 |
| 042 |
|
|
|a dc
|
| 100 |
1 |
0 |
|a Goy, Alexandre Sydney Robert
|e author
|
| 100 |
1 |
0 |
|a Massachusetts Institute of Technology. Department of Mechanical Engineering
|e contributor
|
| 100 |
1 |
0 |
|a Singapore-MIT Alliance in Research and Technology
|q (SMART)
|e contributor
|
| 700 |
1 |
0 |
|a Arthur, Kwabena K.
|e author
|
| 700 |
1 |
0 |
|a Li, Shuai
|e author
|
| 700 |
1 |
0 |
|a Barbastathis, George
|e author
|
| 245 |
0 |
0 |
|a The importance of physical pre-processors for quantitative phase retrieval under extremely low photon counts
|
| 260 |
|
|
|b SPIE,
|c 2021-12-13T16:10:07Z.
|
| 856 |
|
|
|z Get fulltext
|u https://hdl.handle.net/1721.1/136711.2
|
| 520 |
|
|
|a © 2019 SPIE. In a recent paper [Goy et al., Phys. Rev. Lett. 121, 243902, 2018], we showed that deep neural networks (DNNs) are very efficient solvers for phase retrieval problems, especially when the photon budget is limited. However, the performance of the DNN is strongly conditioned by a preprocessing step that consists in producing a proper initial guess. In this paper, we study the influence of the preprocessing in more details, in particular the choice of the preprocessing operator. We also empirically demonstrate that, for a DenseNet architecture, the performance of the DNN increases with the number of layers up to a point after which it saturates.
|
| 520 |
|
|
|a Intelligence Advanced Research Projects Activity (Grant FA8650-17-C-9113)
|
| 546 |
|
|
|a en
|
| 655 |
7 |
|
|a Article
|
| 773 |
|
|
|t 10.1117/12.2513314
|
| 773 |
|
|
|t Progress in Biomedical Optics and Imaging - Proceedings of SPIE
|
