Application de l’Algorithme K-S et de l’Ondelette discrète au traitement d’images satellitaires.

Abstract

This memory approaches the satellite image processing which become a discipline with share today. Its importance is particularly distinguished in case of the forest fires detection, however without doubt; one often faces serious difficulties mainly due to the size of the image, especially when its matrix is treated in hole. To overcome these constraints, an hybrid line by line treatment method is suggested. The latter is based on the algorithm of Kolmogorov-Smirnov as well as the discrete wavelet, which allows obtaining a lossy compression relative to each line of the matrix. In contrast, the algorithm of Kolmogorov-Smirnov permits to detect the stationary segments for each line. The study aims to combine these two methods in order to get a better representation as optimal as possible of the real image.