Shape and topology preserving grey-value pyramids for multiresolution skeletonization

Starting from a binary digital image, a multi-valued pyramid is built and suitably treated, so that shape and topology properties of the pattern are preserved satisfactorily at all resolution levels. The multi-valued pyramid is then used as input data to a grey-level skeletonization algorithm. In this way, a multi-resolution skeleton is computed, which could be useful in many image analysis tasks.