Skip to content
/ DCOD Public
forked from farhadnkm/DCOD

Reconstruct in-line holograms with a single image.

License

Notifications You must be signed in to change notification settings

ni-chen/DCOD

 
 

Repository files navigation

Deep Compressive Object Decoder (DCOD)

Reconstruct in-line holograms merely with a single image. In the proposed model, a regularized untrained deep neural network inversely generates and optimizes the object field based on the acquired hologram.

Check out our paper and results: https://www.nature.com/articles/s41598-021-90312-5

All experiments can be found in the DCOD_Implementation.ipynb

Requirements

To make the workflow expandable and easy to implement, holographic reconstruction algorithms and data processing tools are encapsulated in a python package named Fringe (https://github.com/farhadnkm/Fringe.Py).

To install this package, run:

pip install fringe

This project also requires tensorflow-addons.

Notice:

If you are getting error on hologram import due to LWZ compression, do the following:

pip uninstall tifffile
pip install imagecodecs imagecodecs-lite
pip install tifffile

About

Reconstruct in-line holograms with a single image.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Languages

  • Jupyter Notebook 99.3%
  • Python 0.7%