Research Of Image Segmentation Based On Active Contour Model

With the development of computer technology,digital image processing plays an indispensable role in the information society.In digital image processing,image segmentation,as a key and basic part,has been more and more important in the actual production and life.The purpose of image segmentation is to divide an image into different sub regions according to different features,and separate the interested objects from the background.At present,scholars have put forward many effective image segmentation algorithms.However,image segmentation is still a challenging task due to the problems of intensity inhomogeneity,noise and complex background.The active contour model based on level set can be used in the field of image segmentation because of its ability to deal with the change of topological structure,the simple realization of the numerical method,and the complete segmentation of the object.In this paper,we focus on the problems of intensity inhomogeneity,noise and initial contour sensitivity in the image segmentation,On the basis of LGDF(Local Gaussian Distribution Fitting)model to improve,and achieved the following research results:(1)An active contour model based on multi-scale local features(MS LGDF)is proposed to deal with the problem of intensity inhomogeneity and initial contour sensitivity.First of all,intensity inhomogeneity caused by non-uniform illumination slow changes in the circular smooth scattering shape,the model uses the circular area to get more local information.Secondly,taking into account the intensity varies in different levels in local region,the combination of multi-scale structure and average filter is applied to acquire multi-scale local intensity information.Finally,an approximation of true image,which is obtained by transforming intensity inhomogeneity model,is incorporated into the Local Gaussian Distribution Fitting model and the energy function is constructed with multi-scale local feature.Compared with the traditional LGDF model,the MS_LGDF model has the advantages of high speed,high precision and strong robustness to the initial contour due to the weakening of the intensity inhomogeneity.(2)An active contour model based on local robust statistics(LRS_LGDF)is proposed to deal with the problem of intensity inhomogeneity and noise.First of all,using pixels in the local region of the four quartile range(IQR),the mean absolute deviation(MAD)and median(MED)to construct robust statistical information,what is more,IQR and MAD is the role of sharpening the target boundary,so as to speed up the evolution contour,MED is used to reduce image noise.Secondly,the combination of LGDF and local robust statistics constitute a new data fitting term.Finally,the new date fitting term and the other two internal terms integrate into the variational level set method to construct LRS LGDF active contour model.Compared with the traditional LGDF model,the improved LRS_LGDF model proposed in this paper has the advantages of high speed,high precision and strong robustness to noise due to the introduction of local robust statistical information.