Research On Image Segmentation Method Based On Partial Differential Equations

The traditional Chan-Vese model(C-V model)is not good for image segmentation with uneven gray distribution.The traditional local binary fitting model(LBF)is greatly affected by the position of the initial contour curve,so it is often unable to segment the non-uniform target to the side of the active contour curve accurately.According to the above problems,the traditional C-V model and LBF model are improved in this paper.The main research contents are as follows:(1)The paper improves the local variables based on the traditional C-V model.The improved model can not only accurately segment the images with uneven gray level,but also better describe the gray level changes in the neighborhood of each point,so that the boundary information can be segmented well.Experimental results show that the improved model is more effective than the original one in image segmentation.(2)Aiming at the shortcoming of LBF model in two aspects of segmentation efficiency and accuracy,the paper proposes improved Heaviside function and Dirac function of LBF model.A new regularization function of Heaviside function is selected to make the approximation effect better and improve the accuracy of image segmentation.The new regularization function is obtained by replacing the variables in Dirac function with positive real numbers,which makes the calculation of the model simpler and improves the efficiency of image segmentation.Experimental results show that the improved LBF image segmentation model in this paper has a better image segmentation effect.(3)The traditional C-V model can not segment images with uneven gray distribution well in image segmentation.The LBF model solves this problem well,but the LBF model itself only considers the local information in the segmentation,which is strongly dependent on the position and size of the initial contour curve,so it is difficult to obtain the ideal effect in the image segmentation.So for image segmentation can not only consider the image of local information or just for the global information of the image,so this article about the problem,an improved local segmentation model,on the basis of the global information to join,make when image segmentation curve evolution can be based on local information and global information these two aspects.Experimental results show that the improved model combines the advantages of the two models to achieve higher segmentation accuracy.