Research On Segmentation Method Based On Global Information And Local Information Of Image

Image segmentation,refers to a process or technique that divides an image into two types of object and background or extract objects of interest according to image feature.Image segmentation is a necessary step of image processing,and is an important prerequisite for image analysis.The application of partial differential equations is a great research progress in the field of image segmentation.This breakthrough has made image segmentation technique have a complete mathematic theoretical foundation,and the movement of the evolution curve is more flexible.The active contour model based on partial differential equations,in general,which finds the target contour boundary by minimizing the energy functional of the closed curve,and solves the minimization problem by variational level set method.Based on the above theory,this paper proposes a selective segmentation model based on region information,it takes into account global information and local information of the image,which can selectively segment some relatively complex images.And the model is extended from grayscale images to color images,it solves the problem of selective segmentation of color images.Eventually,the color image segmentation is expressed as the form of dual level set,so that it can segment all objects in the image while segment a single target.For the problem of original selective segmentation that not work well for intensity inhomogeneity images and noisy images,the paper proposes a selective segmentation model based on region information of image.The new model takes into account image region information,combines global region information with local region information.And using a weight coefficient to adjust the ratio of global term and local term when it deal with different types of images.Compared with other models,the simulation results show that the improved model can effectively segment the images and overcome the insufficiency of the original model.The improved model is only for selective segmentation of grayscales images.On the basis of selective segmentation of grayscale images,the segmentation method is extended to the color images.Using the infinitesimal element for arc length of curved as entry points to rephrase the edge stop function of color images,making the function include the information of each component image for color images.Compared with other models,the simulation experiment show that the new selective segmentation model for color images has an ideal segmentation effect on intensity inhomogeneity images and noise images.Based on improved selective segmentation model for color images,we apply the model to dual level set method.There are two level set functions defined,one for selective segmentation by using constraint condition,and the other for segmenting of all objects in the image.The color image segmentation model of dual level set satisfies the higher requirements of people,and the method can realize the selective segmentation of a single target of the color image as well as realize synchronous segmentation of all objects.