On The Study Of Image Segmentation Methods Based On Variational Model And PDE

Image segmentation is a technology of low level image processing, whose result will have an important and direct effect on image analysis and image recognition. In the past several decades, there have been thousands of methods of image segmentation. However, none of them is suitable of all images. In recent decades, there have been a popular method, which is image segmentation based on variational formulation. Its fundamental idea is to change image segmentation into solution of PDE (Partial Differential Equation) so that we can benefit from excellent theory of PDE. Then, more and more people have researched in image segmentation based on variational formulation.In the paper, the applications of PDE in image segmentation will be discussed. The works are as follows:Firstly, the theory of curve has been discussed because the paper has used it in order to express the ACM (Active Contour Model). In the model, the ways of curve presentation has been discussed, and their merits and demerits have been analyzed. Especially, we have emphasis on the skills of curve evolvement process.Secondly, the parameter of ACM has been discussed. The paper has summarized the development of parameter ACM. Based on the Xu's model, GVF, IGVF model has been presented, which improves the convergence of the thin concave boundary where Xu's model has poor convergence.Thirdly, the principle of curve motion of variational model has been unified at first. Meanwhile, the variational models have been analyzed under the guidance of principle.