An efficient active contour with Gaussian distribution fitting energy

Abstract

In this paper, an efficient active contour with Gaussian distribution fitting energy is proposed for image segmentation. We introduce the Gaussian distribution to construct the fitting energy, which drives the active contour toward the object boundaries. In order to protect the smoothness property and the stability, a reaction diffusion approach is borrowed to design a two-stage scheme, which eliminates the procedure of re-initialization of traditional level set methods and improves the segmentation quality effectively.