International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
  
  
  
  
  
Figure 15. Classified image 
The variational algorithm convergences always and rapidly to 
the global minimum, this is not the case for Markov models. 
Moreover, this method showed its performance to classify 
correctly an image after its restoration. We can see in the figure 
9 and 15 that the contours of objects as well as small structures 
are well preserved. : 
8. CONCLUSION 
In this paper, our first objective was to develop a robust model 
for remote sensing image classification. Because images are 
often corrupted with additive noise, we opted for a model that 
combines image classification and edge preserving restoration. 
The edge preserving restoration is not a fortuitous choice, 
because the boundaries information is important for 
classification process. The method we have developed is based 
on the use of van der walls theory in mechanic of fluid, and the 
gamma convergence theory, and consists in construction and 
minimization of a sequence of functional. To avoid smoothing 
on object contours, the regularization chosen is the «9 function 
one. 
The minimization of the functional is achieved by resolution of 
Partial Differential Equations and the use of descent gradient 
algorithm. The implementation of the algorithm showed that the 
method is effective; the image is well restored and classified in 
few iterations. 
To implement the algorithm, we assumed that the image is only 
corrupted with additive noise. To complete this work, we 
project to take in consideration the effect of blur. 
REFERENCES: 
Aubert, G. and Kornprobst, P., 2002. Mathematical problems 
in Image Processing, Partial Differential Equations and the 
calculus of variation. Spring-Verlag, New york, pp. 210-227 
Aubert, G., Blanc-Feraud, L. and March, R., 2002. T- 
Convergence of discrete functionnals with non convex 
perturbation for image classification. Research Report INRIA, 
N° 4560. 
Bertalmio, M., Caselles, V., Rougé, B. and Sole, A., 2003. TV 
based image restoration with local constraints. Journal of 
scientific computing, Vol 19, N°1-3, pp95-122. 
Deriche, R. and Faugeras, O., 1995. Les EDP en traitement des 
images et vision par ordinateur. Research Report INRIA, 
N°2697. 
Mumford, D. and Shah, J., 1989. Optimal approximations by 
piecewise smooth functions and associated variational 
problems. Comm. Pue Appl. Math., 17, pp 577-685. 
Pony, O., Descombes, X., Zerubia, J., 2000. Classification 
d'images satellitaires hyperspectrales en zone rurale et 
périurbaine. Research Report INRIA N°4008. 
Rudin, L., and Osher, S., 1994.Total variation based image 
restoration wih free local constraints. Proc. Of the IEEE ICIP- 
94, vol. 1, Austin, TX, pp. 31-35 
Samson, C., Blanc-Feraud, B., Aubert, G. and Zerubia, J.,2000. 
A variational model for image classification and restoration. 
IEEE transactions on pattern analysis and machine intelligence, 
Vol. 22, N? 5, pp 460-471 
Vogel, C.R., and Oman, M. E., 1996. Iterative methods for total 
variations denoising, SIAM J.Sci Computing, 17(1), pp 227-23 8. 
1214