Research On Active Contour Model Based On Edge Information

With the rapid development of computer science and technology,digital image has been widely used in various fields of production and life.And it has become an important way for people to obtain information.How to extract the information of interest in the image is a hot issue in the field of digital image processing.Image segmentation is the precondition of image classification and recognition.It is impossible to get a good.result without the correct image segmentation.In recent years,image segmentation based on active contour model has become a hot topic in the field of digital image processing.Compared with the traditional segm entation method,it has many advantages.The evolution curve can change the topology raturally.The problem of curve evolution can be transformed into a problem of solving partial differential equations.The main work of this paper is to improve the method of image segmentation based on active contour model.The main work is reflected in the following four aspects:(1)A new boundary indication function is used to improve the effect of the adaptive Distance Regularized Level Set model.In Distance Regularized Level Set model,if the noise of the image is too large,it is easy to make the evolution curve stay in the position of the noise.In addition,it has a bad segmentation result for different types of real images,such as multitargets,weak edges.The diffusion coefficient of the nonlinear diffusion filter is adjusted by the gradient of an image to realize a well-done protection for the edges.The boundary indication function based on the nonlinear diffusion filter is used instead of the boundary indication function in the original model,which reduces the influence of noise,accelerates the segmentation and enhances the ability to detect the weak edge.(2)A new potential well function is used to replace the double well potential function of Distance Regularized Level Set model.The new potential well function reduces the evolution speed near the zero potential well and overcomes the problem of weak boundary.At the vicinity of one well,it is steep,so that the regular term can preserve the character of the distance function.(3)The new gray level information is used to improve the effect of the adaptive Distance Regularized Level Set model.According to the gray scale mean and standard deviation of the image,a new variable weight coefficient is introduced in this paper.This method controls the direction of the evolution curve by the difference between the target and the background,and controls the velocity of the evolution curve by the standard deviation of the image.(4)Based on the information of gradient and grayscale,an adaptive Distance Regularized Level Set model is proposed.The adaptive boundary indication function based on nonlinear diffusion filtering accelerates the evolution rate and avoids the phenomenon of edge leakage.The new area term variable weighting factor can adaptively change the symbol and size according to the image information.It has the opposite sign on both sides of the object boundary,whichenables the zero level set to adaptively select the evolution direction and accelerate the evolution according to the nature of the image.At the same time,it overcomes the dependence on the initial position of the evolution curve,so that the initial curve can be defined at different positions.In addition,the gradient information is used to balance the error caused by relying only on the gray information,andto speed up the evolution of the curve.The experimental results show that the model in this paper can define the initial curve at different positions,which leads to a better robustness to the initial conditions of the level set.At the same time,it has a better segmentation result for different types of real images,such as multitargets,weak edges,multinoise,uneven gray scale.