previousback to the demosnext

Proposed model


 

 We propose to minimize the following set of functionals [7] :
 

 
 
as
 
With
 - the solution we are looking for
-   are the observed data
- is a regularizing function
- W is a potential of level constraint which restricts the admissible values of the solution  to the space of label values.

Roughly speaking, the convergence on could be interpreted as : when is high, only the regularization term is taken into account. As decreases, the regularization vanishes while rising the level constraint term, forcing the solution to take a finite number of values (one of the M label values ).

From a theoretical point of vue, is a Gamma-convergence [6] parameter introduced through the mathematical study of Cahn-Hilliard mechanical models.


previousback to the demosnext
- Introduction
- Algorithm
- Comparison with a stochastic model
- Results : synthetic image
- Results : first satellite image
- Results : second satellite image
- Conclusion

- Bibliography