image_restoration_AI_ZJU

Zhejiang University Course: Artificial Intelligence (19 Spring) Individual Project -- Image Restoration.
Course website is here.

This project uses Python 3.6.

Discription

The task is to build regression models to reconstruct corrupted images. Input images are masked by random noise with specified noise rate.

I implemented two regression models:

• Linear Model with Gassian Basis Function (and analysis by line, as specified in assignment instruction)
• k-Nearest Neighbor Model (which performs better)

Results

0.8 noise (Corrupted/Gaussian Basis/kNN)

0.4 noise (Corrupted/Gaussian Basis/kNN)

0.6 noise (Corrupted/Gaussian Basis/kNN)

0.7 noise (Original/Corrupted/Gaussian Basis/kNN)

0.5 noise (Original/Corrupted/Gaussian Basis/kNN)