Boundary Extraction Using Region-Based GVF Snake Model

Traditional GVF snake model has enlarged capture range and improved the convergence capability for boundary concavities, but it cannot efficiently solve the convergence problem for an image with deep boundary concavities and high noise. In this paper, by integrating the region force derived from the region information of interested object in an image into the force balance equation, a novel snake model was proposed to guide the evolvement of contour curve. Compared with traditional GVF snake model, our model further effectively improves capability of reducing noise sensitivity and converging into deeply boundary concavities. Experimental results with synthetic images and real images demonstrated the feasibility and robustness of our model.