Research On Image Segmentation Model Based On Active Contour

As an important image technology,image segmentation not only obtained the widespread attention,also got a lot of application in practice.In recent years,image segmentation technology based on partial differential equation has become a new cross subject.It has been closely attention because of the relatively mature theoretical system and the more flexible numerical methods.Its basic idea is: using the geometric feature of the image to establish an energy function,and minimizing energy function by the variation method.Finally,we can extract the interest region of image.In this paper,firstly,we introduce the research status and research purposes of image segmentation.Secondly,we introduce and derive the image segmentation theory based on partial differential equation.Then,we analysis the advantage and disadvantage of traditional models and its scope of application in detailed.Finally,we try to improve the existing problems(high computational complexity,sensitivity to initial contour position,poor segmentation effect),and making the model can better segment heterogeneous images with low contrast and complex background.Specific work is as follows:We proposed an adaptive mixture model synthesizing the global information and the local information by a new adaptive balance function.The innovative points of the model are as follows:(1)In view of different image characteristic,the proposed model can adaptively adjust the weighting to drive the curve evolution trend and state.In this way,the intensity information of weak boundaries and complex background can be extracted more precisely,so the model can deal with the image better with low-contrast and complex structure.(2)We add a Gaussian filtering process into model to smooth and standardize the level set function,as well as introducing a parameter to speed up the curve evolution.(3)A penalty term is utilized to eliminate the complicated re-initialization procedure.Experimental results on different kinds of images efficiently demonstrate the good performance of the proposed model in terms of both the speed and accuracy.We proposed a medical image segmentation model based on the combination of edge and region information.In view of the medical image with low-contrast and complex structure,which makes the traditional model cannot deal with the medical image.We know that the edge guide function can accurately extract the image edge,global information can effectively segment the weak boundary and noise image,local information be able to deal with the heterogeneous image better.Based on the advantage,we synthesize three types of information into a unified frame.The innovative points of the model are as follows:(1)The model introduce fitting item of the local information into SPF function of the SBGFRLS model.(2)Using the adaptive weighting function to adjust the local information and global information of the model.Then the evolution speed is faster and the adaptability of the model is enhanced.(3)In the process of numerical implementation,we simplify the model greatly,and reducing the amount of calculation.(4)Finally,A penalty term is utilized to replace the removed term in order to ensure stable evolution of the contours.The experimental results verify the validity of the proposed model,and it can achieve the satisfying segmentation results for the medical images.