R-Snake:A Snake Model Using Both Boundary and Region Information
Yangguang Sun · Dianzi xuebao · 2009
Traditional Snake model is sensitive to the initialization of contour,easily relapsed into a local optimal in a high noise image,and invalid for the image contour with deeply narrow concavities.By designing the image transform operator to derive the region force from the region information included in the interested object,and using Green formula with the conversation ability between region integral and curve integral,a novel snake model R-Snake(Region Snake) was proposed in this paper to extract the contour of interested object,which more directly introduces region information to active contour model in terms of the force balance equation.Because of evolving the contour curve by using both region information and gradient information,our proposed method could not only extend the initialization of contour and alleviate the sensitivity to image noise,but also improve the capacity to converge into complex boundary.Compared with the traditional Snake model,experimental results demonstrated its feasibility and robustness, especially for the images with high noise and deeply narrow concavities.