Skip to content
CNN based non-blind deconvolution (presented at ICCP 2017)
C++ Python Cuda CMake MATLAB Makefile Other
Branch: master
Clone or download
Latest commit d02e57e Nov 3, 2017
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
caffe initial commit Mar 10, 2017
docs/images Add images for a project description Nov 3, 2017
images initial commit Mar 10, 2017
liuetal initial commit Mar 10, 2017
model initial commit Mar 10, 2017
.gitignore modify README.md Mar 10, 2017
LICENSE Initial commit Mar 10, 2017
README.md Update README.md Nov 3, 2017
deconv_cnn.m initial commit Mar 10, 2017
demo.m initial commit Mar 10, 2017
demo_for_cropped_convolution.m initial commit Mar 10, 2017
postprocessing.m initial commit Mar 10, 2017

README.md

Introduction

Unoptimized implementation of "Fast Non-blind Deconvolution via Regularized Residual Networks with Long/Short Skip-Connections" in Matlab and Caffe.

This project uses a Convolutional Neural Network (CNN) to improve a performance of non-blind deconvolution. It uses Wiener deconvolution as a pre-deconvolution, and it enables the network to process a general non-blind deconvolution problem by training various blur kernels and noise levels.

For more details regarding this technique, please refer to the paper

  • Example Example
     * It requires a blur kernel for a blurred image

  • Framework Framework

How to use it

Requirement: Matlab, CUDA<=7.5, Caffe
Tested OS: Ubuntu 14.04
Installation procedure:

  1. build caffe
    $ cd caffe
    -modify parameters and paths in Makefile.config
    $ make all
    $ make matcaffe
  2. set the path of matcaffe in demo.m
  3. run demo.m
  • This code includes only a test function.

Contributors

Hyeongseok Son (sonhs@postech.ac.kr)

Citation

Cite our papers if you find this software useful.

  1. Hyeongseok Son, Seungyong Lee, "Fast Non-blind Deconvolution via Regularized Residual Networks with Long/Short Skip-Connections", IEEE International Conference on Computational Photography (ICCP) 2017, 2017.

About Coupe Project

Project ‘COUPE’ aims to develop software that evaluates and improves the quality of images and videos based on big visual data. To achieve the goal, we extract sharpness, color, composition features from images and develop technologies for restoring and improving by using it. In addition, personalization technology through user preference analysis is under study.

Please checkout out other Coupe repositories in our Posgraph github organization.

Coupe Project

You can’t perform that action at this time.