Abstract—One of the main goals of remote sensing is extracting meaningful information of the image. We are interested in one of the treatments used: segmentation. Thus, we tried to introduce the techniques of learning SVMs (separators Wide Margin) in the field of remote sensing. The SVMs, are tools designed to solve the problems of binary classification supervised, and then generalized approaches (one against one, one against all) classification multi-classes. According to the severability of data, SVMs are distinguished also by two models linear and nonlinear. So we made a classification segmentation of the contents of a satellite image representing the western region of Oran by the separator wide margin non-linear approach and one against all this by using a Gaussian kernel.

Index Terms—Remote sensing, segmentation, separators wide margin.

¹Electronic Department, faculty of Technology, university of Bejaia(06000), Algeria(email : abdelkrim_khireddine@hotmail.com).
²LMOPS Laboratory, university of Metz, 56013, France(email : J.PSalvestrini @univ-metz.fr).

Cite: A.Khireddine and J.p.Salvestrini, "Contribution Separators Wide Margin Processing of Remote Sensing Data," International Journal of Computer and Electrical Engineering vol. 2, no. 4, pp. 778-780, 2010.