Finding the Skeleton of 2D Shape and Contours: Implementation of Hamilton-Jacobi Skeleton

This paper presents the details of flux-ordered thinning algorithm, which we refer to as Hamilton-Jacobi Skeleton (HJS). It computes skeleton any binary 2D shape. is based on observation that points have low average outward flux gradient distance transform. The algorithm starts by computing function and approximating values for all pixels inside Then a procedure called homotopy preserving iteratively removes with high while In this paper, implement transform using fast sweeping algorithm. We present numerical experiments show performance HJS applied various shapes. point out serves multi-scale shape representation, classifier, deficiency detector also quantitatively evaluate reconstructed from medial axis obtained HJS.