一种改善的cv 主动轮廓模型.pdfVIP

To appear in Opto-Electronic Engineering （080491）稿件编号： 一种改进的 C-V 主动轮廓模型 张开华，周文罡，张振，郑孝娟 ( 中国科学技术大学 电子工程与信息科学系，安徽合肥，230027) 摘要：本文对 C-V 主动轮廓模型进行改进，提出一种常微分方程（ODE ）类型的模型。该模型水平集函数采用隐 式方法数值化，避免显式方法数值不稳定的缺点。与传统 C-V 模型相比，不但具有其特点，如可以自动检测带孔 目标的内轮廓等，而且具有以下优点：抗噪性能较优；可快速计算出全局最优分割；水平集方程不含曲率及其它 复杂的差分项，因此可以利用简单的有限差分规则数值化；水平集函数可初始化为零且不用重新初始化；远离轮 廓的边界可以被准确检测；时间迭代步长不受限制。对合成和真实图像的分割结果证明了本文模型具有稳健、快 速的优点。 关键词：图像分割；Mumford-Shah 模型；水平集方法；主动轮廓模型 中图分类号： TP391 文献标志码：A An Improved C-V Active Contour Model Zhang Kai-hua, Zhou Wen-gang, Zhang Zhen, Zheng Xiao-juan (Dept. of Electronic Engineering and Information Science , University of Science and Technology of China, Hefei 230027, China) Abstract : An improved C-V active contour model was presented, with which the level set equation can be solved in an implicit manner. Distinguished from original C-V model, the level set functional of the proposed model is ordinary differential equation (ODE). the proposed model not only has the original C-V model’s characteristics, such as the ability of detecting inner edge of the object with inner hole, etc, also has many advantages as follows: less sensibility of noise; the efficiency of global segmentation is dramatically improved; the initialization of level set functional is very simple, even can be zero, and re-initialization is unnecessary; the edges far away from the active contours would be seriously suppressed by setting the dirac function to be one; without mean curvature and other complicate difference items in the Level set evolution equation, the implementation is much more easier by using a simple finite difference scheme and does not require any limitation of time step. The method was vali