End-to-end learning for digital hologram reconstruction

Zhimin Xu*, Si Zuo, Edmund Y. Lam

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

5 Citations (Scopus)
162 Downloads (Pure)

Abstract

Digital holography is a well-known method to perform three-dimensional imaging by recording the light wavefront information originating from the object. Not only the intensity, but also the phase distribution of the wavefront can then be computed from the recorded hologram in the numerical reconstruction process. However, the reconstructions via the traditional methods suffer from various artifacts caused by twin-image, zero-order term, and noise from image sensors. Here we demonstrate that an end-to-end deep neural network (DNN) can learn to perform both intensity and phase recovery directly from an intensity-only hologram. We experimentally show that the artifacts can be effectively suppressed. Meanwhile, our network doesn't need any preprocessing for initialization, and is comparably fast to train and test, in comparison with the recently published learning-based method. In addition, we validate that the performance improvement can be achieved by introducing a prior on sparsity.

Original languageEnglish
Title of host publicationHigh-Speed Biomedical Imaging and Spectroscopy III
Subtitle of host publicationToward Big Data Instrumentation and Management
Volume10505
ISBN (Electronic)9781510614956
DOIs
Publication statusPublished - 1 Jan 2018
MoE publication typeA4 Article in a conference publication
EventHigh-Speed Biomedical Imaging and Spectroscopy: Toward Big Data Instrumentation and Management - San Francisco, United States
Duration: 29 Jan 201830 Jan 2018
Conference number: 3

Conference

ConferenceHigh-Speed Biomedical Imaging and Spectroscopy
CountryUnited States
CitySan Francisco
Period29/01/201830/01/2018

Keywords

  • Computational imaging
  • Deep learning
  • Deep neural network
  • Digital holography
  • Phase retrieval

Fingerprint

Dive into the research topics of 'End-to-end learning for digital hologram reconstruction'. Together they form a unique fingerprint.

Cite this