Offical PyTorch code for "Photon-Limited Blind Deconvolution Using Unsupervised Iterative Kernel Estimation"

Instructions

1. Create a local copy of repository using the following commands

   foor@bar:~$ git clone https://github.com/sanghviyashiitb/photon-limited-blind.git
   foor@bar:~$ cd photon-limited-blind
   foor@bar:~/photon-limited-blind$ 
   

2. Download the pretrained models, i.e. denoiser and p4ip into model_zoo from the link here

3. Download the zip file containing Levin dataset into the main directory and unzip using the following command:

   foo@bar:~/photon-limited-blind$ unzip levin.zip -d data/ 

4. To test the network using real data, run the file

   foo@bar:~/poisson-deblurring$ python3 demo_deconv.py  

   

Citation

@ARTICLE{9982288,
 author={Sanghvi, Yash and Gnanasambandam, Abhiram and Mao, Zhiyuan and Chan, Stanley H.},
 journal={IEEE Transactions on Computational Imaging}, 
 title={Photon-Limited Blind Deconvolution Using Unsupervised Iterative Kernel Estimation}, 
 year={2022},
 volume={8},
 number={},
 pages={1051-1062},
 doi={10.1109/TCI.2022.3226947}}

Feel free to ask your questions/share your feedback at [email protected]