PRIDA is developed by the lab of computer vision in University of Wisconsin Madison. It stands for Provably Robust Image Deconvolution Algorithm, a image deblurring algorithm.
PRIDA is similar in spirit to the MD algorithm in Convex Optimization. The main difference between the standard MD algorithm and PRIDA is that the step size is chosen independently for each coordinate.
This code is a C++ realization of PRIDA. The matlab implementation is recorded in PRIDA The paper is recorded in arxiv.
CMake should be installed
OpenCV should be installed -> Install-OpenCV
git clone https://github.com/tianyishan/PRIDA_CPP.git
cd PRIDA_CPP
mkdir build
cd build
cmake ..
make
The format of command-line argument is:
./prida <imagename>.png or pathname lambda kernel_size
To run one of the demo pictures:
./prida ../babies.png 0.0006 19
When the program finishes, it will write the result into the same folder of your input image.
The future goals of this project are
- To optimize the speed of running the algorithm.
- To adapt it with modern deep learning methods.
- To work with a wider variety of blur kernels and images.
- To create a GUI.