Poisson Skeleton Revisited: a New Mathematical Perspective
Journal of Mathematical Imaging and Vision
| Hi-index | 0.00 |
The problem of computation of a "good" skeleton is still open because several somehow opposite requirements must be satisfied: the skeleton must represent the connected components of the shape (connectivity requirement). The skeleton must be noise insensitive. The computation must be as independent as possible of the grid effects. We discuss several classical "thinning" algorithms and show that they can be reinterpreted as partial differential equations governing the shape evolution. We propose the best adapted partial differential equation to the computation of the skeleton, and define a reliable numerical scheme to compute it. Experiments and comparison of methods close the paper.