Improved Image Segmentation Model And Fast Algorithm Implementation Based On Euler Elastic Energy

In this paper,the image segmentation model with Euler's elastica energy and the numerical implementation of the augmented Lagrange algorithm are mainly studied.The case that target boundary is broken,blurred or partially covered makes the segmentation difficult.In view of this situation,the image segmentation model with Euler's elastica energy as the regularization term figure out this problem effectively,such as ECV-L2 model's automatically connecting broken edge and ECV-L1 model's preference to segment convex contour.However,due to the influence of curvature term of Euler's elastica energy,it is difficult for the level set function to reflect the edge of the targeted object.In order to make the model maintain the properties of Euler's elastica energy and have a segmentation result closer to the edge of the target object.Based on the ECV-L1 model,a new model fusing edge detection function is proposed in this paper.Augmented Lagrange algorithm is used to solve the new model,and the evolution of the level set function in the process of solving the new model is demonstrated in this paper.The accuracy of the segmentation results of the ECV-L1 model and the new model is compared by calculating the similarity function between the segmentation curve and the targeted object in the segmentation process of the ECV-L1 model and the new model.Numerical experiments show that the new model can more accurately reflect the edge of the target object while it tends to segment the convex contour.