Self-tuning-imgrestoration

This a Self-Tuning Image Restoration.

• Extraction of Kernel/PSF used in a JPEG file
• Then use the extracted info as a parameters in a unsupervised wiener algorithm for restoration
• Only works for grayscale images (for now)
• Created a Web interface React (Front End) and Python Flask (Back End)
  • Filter Page
  • Restoration Page

Dependencies

1.) Python handles the image processing part of this system. Dependencies includes below:
PIL, Matplotlib, Numpy, Scipy, Scikit-learn

2.) React framework also need dependencies on the front end.
Piexif-JS, Router-Dom

Instruction

1.) Upload an Image

2.) Select a filter and its strength (supports Box Blur, Sharpen, Gaussian Blur, and Sample Blur only)

3.) Download the filtered Image and you can see the filtered info (its type and kernel used) in its properties.

4.) Upload the saved filtered image in the restoration page

5.) save the restored Image

Comparison

Acknowledgement

Thanks to my Senior Job Lipat for introducing and helping me with React Framework