【作者】 李忠伟； 潘振宽； 倪明玖；

【Author】 Li Zhongwei~1,Pan Zhenkuan~2,Ni Mingjiu~1 (1 College of Physics,Graduate School,Chinese Academy of Science,Beijing,100049; 2 College of Information Engineering,Qingdao University,Qingdao,Shandong,266071)

【机构】 中国科学院研究生院物理学院； 青岛大学信息工程学院；

【摘要】 本文提出一个将图像除噪(TV)模型和图像分割(CT)模型相结合的多相图像分割模型;该模型在分割过程同时完成去噪,缩短分割执行时间,提高分割效率。TV(Total Variation)模型是一种经典的图像处理的变分模型,在图像噪声去除、图像修复、图像分解等方面取得成功应用。该模型的TV规整项使得恢复后的图像保持图像边缘;CV模型是一种基于水平集的图像分割模型;本文将用连续的符号距离函数定义水平集函数,用n个水平集函数划分2~n个区域,用区域编号的二进制数和水平集的Heaviside函数表达通用的区域划分特征函数,从而构造基于CV模型多相图像分割模型。最后把该多相图像分割模型成功应用于合成图像和实际图像的分割。更多还原

【Abstract】 In this paper a new multiphase image segmentation model is proposed that combines the image de-nosing(TV) model with the image segmentation (CT) model.This model can complete the de-noising in the segmentation process. So it can shorten the execution time partition and improve the efficiency of segmentation. TV(Total Variationa) model is a classical variaonal model in image processing and has been widely used in image de-noising,image in-painting and image decomposition.The total variation regularizer in TV model leads to preserving edge.CV model is an image segmentation model based on level set.N level set functions defined as signed continuous distance functions are used to partition 2~n regions; a generic characteristic function formulation of each region is designed using binary numbers of region’s index and Heaviside functions of level set functions. Therefore we propose a multiphase image segmentation model based on CT model. The proposed model has been applied to both synthetic and real images with promising results.更多还原