COMPARISON OF THREE DIFFERENT IMAGE FORCES FOR ACTIVE CONTOURS ON ABDOMINAL IMAGE BOUNDARY DETECTION

Abstract

Active contour, or snake, is an energy minimizing spline that is useful in image boundary detection. Active contours are stimulated by internal forces, image forces and external forces which maintain the shape of the contours while attract the contours to some desired features, usually edges. Problems in implementing active contours such as convergence and initialization have motivated researchers to modify image forces of the active contours. This paper presents a comparative study among three different image forces: traditional snakes, balloon and gradient vector flow (GVF). The study is validated by experiments on abdominal image boundaries detection. These lead to the conclusion that GVF gives the most appropriate results among the other approaches.