Research On Image Segmentation Based On Active Contour Model

In the era of big data,humans have produced a large amount of image data,which contains a huge amount of wealth.In order to effectively extract information from massive image data,digital image processing has received attention from all walks of life and has developed rapidly in the past decade or so.Image segmentation is an important part of digital image processing,but boundary extraction is one of the important branches.There are many kinds of boundary extraction algorithms,of which the active contour model is a kind of boundary extraction algorithm based on differential equations.Because this type of algorithm has the advantage that the segmentation result is a closed curve and it is relatively simple to implement,it has become a hot topic in current research.This paper summarizes the basic theory and basic knowledge of the active contour model,and introduces several classic active contour models under the basic theoretical framework.Through combing the active contour model,it is found that there are still many problems to be improved in the active contour model,including the traditional computation of the active contour model with high computational cost and low noise immunity.It is difficult to segment the complex image effectively.In order to improve the computational efficiency of the traditional active contour model and the ability to segment complex images,this paper studies the image segmentation algorithm based on active contour model.The main contributions of this article are as follows:First of all,for the problem of large amount of computation and long time required to calculate the vector field by using the diffusion in the gradient vector flow active contour model,a new calculation method is proposed—partial diffusion partial interpolation.The method first calculates the external force field of GVF by using the diffusion method on some pixels of the image,and then uses the interpolation method to calculate the external force vector field of the remaining points.The use of partial diffusion partial interpolation method can save the calculation and calculation time,and to a certain extent can enhance the anti-noise ability.Secondly,this paper studies the saliency model and the active contour model,combines the two effectively and proposes a saliency-based active contour model algorithm.The algorithm uses the vector field generated by the saliency map and the gradient vector flow vector field of the original image to evolve a closed curve so that the curve can express the outline of the target.Experiments show that the algorithm obviously improves the ability of traditional gradient vector flow active contour model to segment complex images.Finally,the traditional image segmentation algorithm does not use prior information and only uses the information of the image itself for segmentation.This paper proposes a nonparametric statistical active contour model algorithm that can use the prior information of the distribution.Using distribution prior information can effectively improve the segmentation effect and the segmentation efficiency.In order to improve the speed of operation,this paper also proposes a method to speed up the calculation.The experimental results show that the proposed algorithm can achieve good results for the segmentation of artificial pictures and natural pictures.The comparison experiments show that the proposed algorithm has obvious improvement in accuracy and computational efficiency compared with the traditional nonparametric statistical active contour model without prior information.