Skeletonization of Gray-Tone Images Based on Region Analysis

A problem often present in skeletonization of gray-tone digital images is that the obtained skeleton includes an excessive number of branches. In this respect, a regularization process should be performed in order to partially, or totally, remove branches which are not meaningful in the problem domain. In this paper, we propose a skeletonization algorithm which is active only on a suitable subset of the image, mainly constituted by regions understood as relevant from a perceptual point of view. The notion of dominance of a region, which is defined in terms of geometrical features, gray-value and adjacency relations, plays a central role in the selection of the regions of the subset. The obtained skeleton turns out to be more representative and its simpler structure will allow one to perform the regularization phase with a reduced computational effort.