Previous :
Choice of the estimator Next
: Log-Likelihood derivatives |

Calculus of the Likelihood |

**The different probability laws :**

follows aYGaussiandistribution(),N:N=Y-HX

- The prior is given by a
Gibbsianprobability, becauseis supposed to be a Markov random field.X

** The
likelihood can be rewritten in the form :**

,Z=Z_{Y}partition functions(cf. statistical mechanics) ,

they are normalizing constants.

is the set of images withgrey levels.M

, at least :

Impossible todirectlycalculate the partition functions !

because of the dimension of the space.

We do not try to calculate them, but to estimate them.