Here we use the position of each region inside the image, its size, structure and adjacency relations to the other prominent regions or to the background. First we apply a so-called preprocessing step, in which we separate objectable images from non-objectable ones with the meaning of whether or not the given image contains prominent regions.