Image Segmentation Methods Based On Partial Differential Equation

Image Segmentation is the key step in the Image Processing, there are a lot of methodsfor Image Segmentation. In this paper, the theories of Partial Differential Equation, CurveEvolution and the Level Set method were used in image segmentation. The main work is asfollows:Firstly, a new edge indicator function is presented, which could make the evolutionarycurve quickly converge to image edge on the basis of the model in this article; The statisticalvalues of special gray were used to divide image into two types according to the principledefined by this article, procedures and parameters will be selected during evolutionary processin light of the divided kinds.Secondly, internal and external average gray values of Chan-Vese (C-V) model aremodified in order to embed edge information into C-V model; The new model is constructedby combing the advantages of the edge-based image segmentation model with C-V model,which eliminates the need for re-initialization, could automatically select the curve evolutiondirection of movement, and quickly converge to image edges.Finally, the difference between primary and subordinate level set functions was used tocontrol the iteration times. The segmentation results were considered to be stable if thedifference is small enough, and the iteration is finished. A large number of numericalexperiments were given in this article, and good segmentation results illustrate the superiorityof the edge indicator function and the new model in this article.