Research On 3D Image Segmentation And Surface Denoising Based On Variational Method

With the maturity and development of computer science and technology,mathematical analysis and physics,the application of partial differential equation technology in image processing has gained wide attention at home and abroad.In the field of image processing,image segmentation and image denoising are the foundation and key of image processing technology.Three-dimensional image multi-phase segmentation and three-dimensional surface denoising are the focus and difficulty in this field.At present,3D multiphase segmentation has very important research value in 3D image remodeling,geophysical exploration,pattern recognition and medical diagnosis.Because the error is the inherent existence of data in the process of measurement and transmission,there is inevitably noise phenomenon on the surface of 3D image after segmentation and remodeling of CT and other fault image sequences,which will affect the subsequent research work of the image.Therefore,3D surface denoising is an important part of 3D image processing.Based on 3D image,this paper studies the content of multiphase image segmentation and surface denoising.This paper introduces the variational level set method theory for 3D multi-phase image segmentation,analyzes the characteristics of Chan-Vese image segmentation model.Finally,the 2D image Chung-Vese model is extended to 3D image multi-phase segmentation.Split Bregman projection algorithm combined with variational method is used to solve the optimization problem of energy functional.Finally,the effectiveness and versatility of the model for 3D image 2 phase and multi-phase segmentation are verified by a large number of experiments.The high order model and total generalized variational model based on laplacian operator are extended to surface denoising.The variational method and ADMM fast algorithm are used to solve the energy functional optimization problem.The ideal experimental results are obtained through a large number of simulation experiments.the surface denoising model in this paper is compared with the existing surface denoising models(TV model,MC model and TC model).It is verified that the model can keep the edge profile of the 3d surface image while effectively removing the surface noise.