Image Transmission Through a Dynamically Perturbed Multimode Fiber by Deep Learning

Shachar Resisi, Sebastien M. Popoff, Yaron Bromberg*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

29 Scopus citations


When multimode optical fibers are perturbed, the data that is transmitted through them is scrambled. This presents a major difficulty for many possible applications, such as multimode fiber based telecommunication and endoscopy. To overcome this challenge, a deep learning approach that generalizes over mechanical perturbations is presented. Using this approach, successful reconstruction of the input images from intensity-only measurements of speckle patterns at the output of a 1.5 m-long randomly perturbed multimode fiber is demonstrated. The model's success is explained by hidden correlations in the speckle of random fiber conformations.

Original languageAmerican English
Article number2000553
JournalLaser and Photonics Reviews
Issue number10
StatePublished - Oct 2021

Bibliographical note

Funding Information:
The authors kindly thank Snir Gazit for providing access to the computational resources which were used to train the neural networks, along with Ori Katz and Roy Friedman for many fruitful discussions and suggestions. S.R. and Y.B. acknowledge the support of the Israeli Ministry of Science and Technology and the Zuckerman STEM Leadership Program. S.M.P. was supported by the French Agence Nationale pour la Recherche (grant No. ANR‐16‐CE25‐0008‐01 MOLOTOF), the Labex WIFI (ANR‐10‐LABX‐24, ANR‐10‐IDEX‐0001‐02 PSL*), and the France's Centre National de la Recherche Scientifique (CNRS; France‐Israel grant PRC1672). All authors acknowledge the support of Laboratoire international associé Imaginano.

Publisher Copyright:
© 2021 Wiley-VCH GmbH


  • deep learning
  • endoscopy
  • image reconstruction
  • imaging
  • multimode optical fibers
  • speckle


Dive into the research topics of 'Image Transmission Through a Dynamically Perturbed Multimode Fiber by Deep Learning'. Together they form a unique fingerprint.

Cite this