Photon-Limited Non-Blind Deblurring using Algorithm Unrolling

Yash Sanghvi, Abhiram Gnanasambandam, Stanley H. Chan

PDF

Arxiv

Dataset

Code

Abstract

Image deblurring in photon-limited conditions is ubiquitous in a variety of low-light applications such as photography, microscopy and astronomy. However, the presence of photon shot noise due to low-illumination and/or short exposure makes the deblurring task substantially more challenging than the conventional deblurring problems. In this paper we present an algorithm unrolling approach for the photon-limited deblurring problem by unrolling a Plug-and-Play algorithm for a fixed number of iterations. By introducing a three-operator splitting formation of the Plug-and-Play framework, we obtain a series of differentiable steps which allows the fixed iteration unrolled network to be trained end-to-end. The proposed algorithm demonstrates significantly better image recovery compared to existing state-of-the-art deblurring approaches. We also present a new photon-limited deblurring dataset for evaluating the performance of algorithms.

Video Presentation - ICASSP 2022

Slides

Dataset

Click here for download and further details about the dataset!

Real world dataset for evaluation of non-blind deblurring algorithms in the presence of photon shot noise. Contains 30 images at different light levels and blurred by different motion kernels - ground truth kernel captured using a point source.

Results Gallery

❮ ❯

Citation

 @ARTICLE{9903556,
  author={Sanghvi, Yash and Gnanasambandam, Abhiram and Chan, Stanley H.},
  journal={IEEE Transactions on Computational Imaging}, 
  title={Photon Limited Non-Blind Deblurring Using Algorithm Unrolling}, 
  year={2022},
  volume={8},
  number={},
  pages={851-864},
  doi={10.1109/TCI.2022.3209939}}