Choosing the root node of a quadtree

Granular computing is closely related to the depth of the detail of information with which we are presented, or choose to process. In spatial cognition and image processing such detail is given by the resolution of a picture. The quadtree representation of an image offers a quick look at the image at various stages of granularity, and successive quadtree representations can be used to represent change. In this paper we present a heuristic algorithm to find a root node of a region quadtree which reduces the number of leaves when compared with the standard quadtree decomposition.