Ising/Pott Model    

Applet written by Sylvain Reynal
This applet illustrates the Metropolis sampling method, an algorithm able to sample several realizations of a given Markovian (or Gibbs) distribution. This Markovian/Gibbs distribution corresponds to the simple Ising/Pott model studied in course (see Slide 13) using q class labels, a first order neighborhood with identical potential value for the different cliques (thus a model favoring homogeneous regions such as the one commonly used in a classical Markovian segmentation procedure) for different temperature values (see also Slide 19) and watch the system evolves in real time.

Another interesting applet showing realizations of a given Markovian/Gibbs distribution can be found here