Optimal inference for hierarchical skeleton abstraction

Skeletons are well-known representations that accommodate shape abstraction and qualitative shape matching. However, skeletons are sometimes unstable to compute and sensitive to shape detail, thus making shape abstraction and matching difficult. To address these problems, we propose a principled framework that generates a simplified, abstracted skeleton hierarchy by analyzing the quasi-stable points of a Bayesian-inspired energy function. The resulting model is parameterized by both boundary and internal structure variations corresponding to object scale and abstraction dimensions, and trades-off reconstruction accuracy and representation parsimony. Our experimental results show that the method can produce useful multi-scale skeleton representations at a variety of abstraction levels.