An active contour method for tubular object segmentation

There are many tubular objects in image segmentation problems, such as vessel, road, and river and so on. In this paper, an active contour method is proposed for tubular object segmentation. This method is inspired by capillarity and based on the tubular structure enhancement. We get the tubular structure response direction as well as the response intensity by analyzing the Hessian matrices, and the response result is incorporated in the active contour model to control the curve´s evolution just along the tubular object´s extend direction. This active contour method is fit for tubular object segmentation, the experiments results valid our method.