With the three following "maps", considered as complementary information, we must find how to "compute" the adequacy of each point of the image to the favourable context, for each possible vegetation class.

Note that we must first compute some information such as slope, valleys, ridges, distance to roads or villages and the more complicated "irrigable zone''. This information is called "derived maps". At this step, this "out-image" information may sometimes be extracted by adapted algorithms from the image itself.

For example:

• urban areas,
• roads,
• digital elevation models (from stereo pairs of satellite images)...

In this case, even if no digitized map is available, such information can be produced (even not complete) from the image itself in order to take the expert knowledge into account.

• Digital Elevation Model
• Roads
• Urban Areas and Villages