Segmentation of mechanical images using improved active contour model

An approach to boundaries of mechanical images using active contour model was discussed. Active contour model is used widely in image segmentation. Traditional snake model has poor convergence to concave boundaries. GVF snake model solves the problem, but it has some deficiencies such as it can not capture boundaries like U and bottleneck completely because of the counteraction of some external force and the influence of the local minimum external force. In this paper, an additional control force is added to GVF snake model. The direction of control force is gained by tracking the motion of snake´s control points. Experimentation results show that the improved snake model can solve the problem that GVF snake model can not detect the boundaries like U and bottleneck and the new algorithm can improve GVF snake model´s ability to capture thin boundary indentation.