An Active Contour Algorithm Based On Improved Gradient Vector Flow

Snakes, or active contours, are used extensively in computer vision and image processing applications, particularly to locate boundaries. Compared with traditional active contour models, the Gradient Vector Flow (GVF) method can deal with problems associated with initialization and poor convergence to boundary concavities. In this paper, an active contour algorithm, which based on improved gradient vector flow, is proposed. It retains the improvements of the GVF and overcomes some disadvantages. It also adds the following new features: firstly, it leads to a fair diffusion of the boundary information, and induces the new gradient vector flow the ability to overcome weak edges due to noise presence. Secondly, because of the pace and blending factor, the active contour firstly evolves to the area of strong gradient boundary with high speed, and then evolves to the actual boundary of its neighborhood with low speed, therefore, it does not miss boundary details. Finally, in practical application, the algorithm converges quickly and has less affection of the blending factor.