Research On Method Of Image Segmentation Based On Regional Active Contour Model

Image segmentation is an extremely important research topic in the field of digital image processing and computer vision.It is a process that divides images into specific and unique regions and extracts the interest objects.It is widely used in medical treatment,remote sensing,transportation and other fields,bringing great convenience to people's production and life.However,due to the complexity of the image,and there are many kinds of images.So far,there is no universal image segmentation method.Nowadays,for the segmentation of images,many scholars have proposed different methods.Among them,the active contour model is supported by flexible topological transformation and strong mathematical theory,making it a hot research topic in the current research field.This paper mainly studies the image segmentation method based on the regional active contour model,which is an image segmentation method based on the curve evolution theory and the level set method.Based on the image segmentation method based on the regional active contour model.Aiming at the problem that the model is sensitive to the initial contour,and has boundary leakage or mis-segmentation in the segmentation of weak edge image and intensity inhomogeneous image,resulting in the contour curve can not be well evolved to the obj ect boundary.The method based on regional active contour model is studied,improved and implemented in this paper.The main work is as follows:(1)As active contour model is sensitive to the initial contour position,cumbersome selection and multiple iterations in image segmentation with intensity inhomogeneity,an active contour model is proposed based on local entropy fitting energy and global information.Firstly,intensity image center is selected as the center point of the level set initial contour to determine initial contour location by changing radius size.Then,local entropy is used to enhance image edge response,which puts it together with Region-Scalable Fitting(RSF)model as local energy terms and adds image global information without falling into local minimum.Finally,regular terms are redefined to improve segmentation efficiency.The model is applied to synthetic and real medical images with intensity inhomogeneity.The proposed method can better deal with the intensity inhomogeneous images.(2)Aiming at the problem that the active contour model can not evolve well to the target boundary in segmenting weak edge images and severe intensity inhomogeneous images,this paper proposed an active contour model based on local enhancement and region fitting.Firstly,the original image is converted to a new image using a local area enhancement method to enhance the contrast of the image.Secondly,the statistical fit is used to calculate the region fitting energy of the image.Then,add a regular term to avoid re-initialization of the evolution contour and improve image segmentation efficiency.Finally,the model is applied to synthetic and real images with intensity inhomogeneity,the experimental results validate the favorable performance of the proposed model.