Abstract
In this paper, an algorithm for thinning of an object from its binary image is described. A generalized potential field is defined over the interior of the object and the divergence of the corresponding vector field is computed for each interior pixel. Then, those pixels of the object whose divergence values are above some specified threshold are selected, resulting in a thinned version of the original object. The density of the resulting skeletal structure can be controlled using different threshold values. The thinned version of the object obtained by this method can be used for image compression and shape matching.