Algorithm for Fast Image Restoration

Blind deconvolution, which comprises simultaneous blur and image estimation, is a strongly ill-posed problem. It is by now well-known that if multiple images of the same scene are acquired, this multichannel blind deconvolution problem is better posed and allows of blur estimation directly from the degrade images. We improve the multichannel idea by adding robustness to noise and stability in the case of large blurs or if the blur size is vastly overestimated. We formulate blind deconvolution as a L1-regularized optimization problem and seek a solution by alternately optimizing with respect to the image and with respect to blurs. Each optimization step is converted to a constrained problem by variable splitting and then addressed with an augmented Lagrangian method, which permits simple and fast implementation in the Fourier domain. Our current implementation can also perform superresolution.

The algorithm is implemented in MATLAB (requires Image Processing Toolbox). Instructions on how to set parameters are in parameters.m. You can download the code here.

We also provide a MATLAB code for our ECCV2012 paper "Deconvolving PSFs for a Better Motion Deblurring Using Multiple Images", which extends the original blind deconvolution algorithm by adding a step of deblurring estimated blurs to reduce spurious blurs that often occur in blind deconvolution. The code can be downloaded here.

By downloading the code you accept the following terms:

  1. I will use the tools solely for educational and/or research purposes.
  2. I will not provide the code to any third party without first asking
    for permission the authors of the tools.
  3. I will acknowledge the work of the authors of the tools in all my
    publications, which will utilize results, ideas or parts of the code
    in some way, and I will send the preliminary or final publications to
    the authors.

Contact Filip Sroubek to inquire about the MATLAB code.

 

The flowchart of the proposed algorithm

 

 

Results: (images were taken in the charming town of Litomerice with DSLR Canon Rebel T2i)

 
 Input out-of-focus images
  

Estimated sharp image and blurs

 
 Input out-of-focus images
 
 
Estimated sharp image and blurs

 

 

Details:  
Duration: 2010-2012
Contact person: Filip Sroubek
Involved people: Peyman Milanfar, Jan Kamenický, Xiang Zhu

Publications:

AttachmentSize
paper published in TIP 20122.51 MB
Supplementary material3.6 MB
MATLAB code for fast image restoration918.15 KB