Research On Segmentation Method Based On Active Contour Model

The image segmentation problem is fundamental in the field of computer vision and image processing. There are lots of algorithms have been proposed to solve this problem after years of research and development.Putting forward active contour model is a breakthrough in the field of segmentation.It is widely used in medical image segmentation, target tracking, edge detection and so on due to that it is based on the rigorous mathematical theory and other prominent advantages.This article is mainly research on the method of the active contour model. Firstly we will classifiy and describe the active contour model, and then introduce some of relevant mathematical theory, and finally get results of those classical active contour model by doing some experiments and make the brief analysis of the advantages and disadvantages of the model according to the results. Through analysis it is found that the SBGRLS model can not deal with the image with intensity inhomogeneity. In view of this, this paper uses the differential image information to reconstruct the signed pressure force function, so as to get a new model based on difference image information.Experimental result show that the proposed model can accurate segmentation intensity inhomogenous images compared with SBGRLS model, and in the computation is more efficient than LBF model.In research of CV model and LBF model found that CV model can not deal with the image with intensity inhomogeneity and the LBF model is sensitive to the initial contour. Therefore, this paper proposed a novel hybrid active contour model. The proposed model consists of three terms a joined the boundary information in the CV model as the global term and LBF model as the local term and curve length term, respectively. In order to avoid the use of traditional level set methods for re-initialization, the model using the binary level set method in the implementation,which not only avoids the re-initialization process and the numerical implementation is more simple. In global term with boundary information which added to the ability to capture the boundary, at the same time using the local information of images, therefore,the model can deal with the image with intensity homogeneity and inhomogeneity.Experimental results show that compared with classic active contour model to get good results in deal with the image with intensity inhomogeneity, also has some improvement in the sensitivity to the initial contour.