Fractional Differentiation-Based Edge Energy Driven Active Contours for Robust Image Segmentation

In this paper, a new fractional differentiation-based active contour model for robust image segmentation is presented. A new edge energy is introduced, in which the contour evolution is driven by the difference between the fractional derivatives directed along the inward and outward normal directions of the evolving contour. We provide the level set formulation of this novel energy and show that this energy is minimized when there is an accurate alignment of the zeroth level set of the evolving contour with the actual object boundary. The proposed model outperforms other state-of-the-art active contour methods in eliciting weak/fuzzy boundaries in real world images and provides robust segmentation even under influence of various types of noise as quantified by segmentation metrics.