Shape description by a distribution function based on skeleton structure

A new method of describing the shape of a silhouette for data compression is proposed. A silhouette is decomposed into a union of components. Each component is defined as a density distribution function. The algorithm of shape decomposition is based on the skeleton structure. The components are represented in a tree structure according to the sequence of silhouettes. The latter are generated by applying a Gaussian filter to the original contour at varying levels. Simulation reveals that the proposed technique requires fewer bits than other shape decomposition schemes at the same degree of approximation