The aim of this chapter is to delineate landscape zones based on raw and high resolution satellite data using standard image processing capabilities (no programming). These landscape zones will be used in further research in order to generalise the local results of a household survey to the regional level (scale transfer) and improve the understanding of agrarian systems in Fouta Djallon using GIS. The method proposed involved a preliminary stage of identification of the optimal resolution in order to extract spatial information from an image using local filtering techniques. Then, the spectral information is synthesised to avoid redundancies. The spatial information is extracted using a standard high pass filter applied after having degraded the data to the optimal resolution. After a generalisation stage, a "pixel per pixel" based clustering is undertaken with, if necessary, location variables. The result consists in the delineation of homogeneous zones on a raw and high resolution satellite image that can be interpreted on the ground as landscape zones.