| Image segmentation is an important research aspect in computer vision. And the goal of Image segmentation is dividing the image into a series of different sub-region according the characteristic of image such as gray, texture, color, optic and so on. As complexity of the image and multi-object recognition, multiphase image segmentation instead of twophase image segmentation will get more research and development in the future image segmentation field. But the drawbacks of local minimization and low efficiency are two problems of multiphase image segmentation based on level set method to limit their applications in different areas. And the improved segmentation models and rapid algorithms are designed in this paper towards two drawbacks. Firstly, the theory and of curve evolution, level set method, variational Chan-Vese Model and their applications are discussed. Secondly, the basic principle of global convex minimization method for two phase image segmentation and rapid Split Bregman algorithms are illuminated. Global convex minimization for the proposed model is implemented by introducing characteristic function replacing traditional level set method to convex relax discrete region to continuous one. And The Split Bregman method is implemented by introducing auxiliary variables which transform the relaxed convex variational model into solving simple Poisson equations and exact soft thresholding formulation. Thirdly, the proposed model can be used for image segmentation of any phase, and can be used for segmentation of 2D and 3D images in the same form, and the multiphase segmentation model for image on implicit surface is also proposed. All of the models can be convex relaxed to a seies of sub-optimization problem solved by Split Bregman algorithm. Some numerical experiments demonstate high efficiency of the proposed model via comparsions with the tradtional method. Finally, future research direction is specific according the drawbacks existing in the paper. |
PDF Full Text Request