Project1:    Artificial Intelligence:  Image Restoration 

Programming Tool:    Java ( JDK 1.2)

Usage: Clicks buttons in order (left to right) 

Objectives

  1. To choose one artificial intelligence algorithms, and write a java applet, compile and incorporate it as part of a web page that can be viewed using a browser.
  2. To animate the steps of the algorithm dynamically, so that other users can understand each step of the algorithm by viewing the java applet.

Image Restoration by MCMC on MRF

For fast implementation, preprocess the image.
1. Re-scale the input image to 5 intensity levels [2,3,4,5,6].
2. Add Gaussian noise N(0, 2^2) at each pixel, truncate the intensity to [0,7].
3. When you display the image on the screen, you should re-scale the output image intensity by 32 times (i.e. back to [0,224]), and you may enlarge the size of the image if necessary.

Using the Monte Carlo Markov chain sampling algorithm discussed in class, your algorithm recovers from the distorted image a "clean" one. Show one final result that you have computed off-line on the screen, and show the process of pixel intensity change on-line. At each pixel, the MCMC algorithm choose between intensity levels [0,1,...7] at local probabilities.