A Variable-Step Detecting Algorithm for Interested Boundary

Traditional detecting algorithm for interested boundary is inefficient and gets too many useless edge points, in order to resolve these problems, a boundary points estimate and search algorithm was proposed benefit from the thoughts of Snake algorithm and boundary tracking algorithm. By analyzing the relationship between curvature distribution and boundary shape, a step size calculating formula was deduced, and a variable-step detecting algorithm based on curvature limitation was advanced. The experiment results show that the distribution of edge points produced with our new algorithm agrees with the boundary shape well, and compared with traditional detecting algorithm, the points number is much fewer and the detecting efficiency is much higher