Interferometric Lensless Imaging

Rank-One Projections of image frequencies with Speckle Illuminations

---

Repository Structure

The repository is organized as follows:

|- images/                      images to be reconstructed.
|- paper_experiments/           code that generated figures 2,3, and 6 of the paper.
|- reconstructions_demo/        some notebooks that analyze the reconstructions dependency on various parameters.
|- tests_and_visualizations/    tests, visualizations, and additional code examples.
|- utils/                       utility functions and modules.
|- demo.ipynb                   Jupyter Notebook demonstrating the project functionalities.

Installation and Usage

To use the code in this repository, follow these steps:

1. Clone this repository using git clone or download the zip file.

2. Create a minimal virtual environment and activate it. We suggest conda, if you want to use pipenv instead, replace step 2. below by the instructions in pipenv.md . With conda, use the following commands:

conda create --name your_env_name python==3.8.8
conda activate your_env_name
pip install -r requirements.txt

This project uses old versions of Python, scipy, and other packages. requirements.txt ensures you will run the code in a controlled environment.

Note: to see the newly create venv, you may need to relaunch your python editor.

To deactivate the environment when you don't want to work on this project:

conda deactivate

3. install my forks of pyunlocbox, pyproximal and pylops from Github:

   • clone each repo from Github or download the zip file

   git clone https://github.com/olivierleblanc/pyunlocbox.git 
   

   • then in command line, ensure the created virtual env is activated, then pip install the repo from the fork you just downloaded.

   pip install -e /path/to/repo
   

   repeat with
   https://github.com/olivierleblanc/pyproximal.git and https://github.com/olivierleblanc/pylops.git

4. Run the demo Jupyter Notebook (demo.ipynb) to see an example of how to use the project functionality.

5. Explore the project files and directories to find the desired code and resources.

Data

The experimental data can be download here

Contributions

Contributions to this project are welcome. If you find any issues or want to add new features, please open an issue or submit a pull request.

License

This project is licensed under the MIT License.

Contact

Don't hesitate to discuss with me eiter on Github directly or email me at [email protected].

Note

In the code, you sometimes used the word diagless or dl that is a synonym of hollow matrix. It is related to the debiasing trick explained in the paper.