基于水平集曲线演化的目标轮廓提取  

Object Contour Extraction Based on Curve Evolution via Level Set

作　　者：陈波[1] 赖剑煌[2] 

机构地区：[1]中山大学数学与计算科学学院,广州510275 [2]中山大学信息科学与技术学院

出　　处：《计算机科学》2006年第8期227-228,235,共3页Computer Science

基　　金：国家自然科学基金(60373082);教育部科学技术重点项目(105134)

摘　　要：基于水平集的曲线演化与活动轮廓模型相结合,产生了几何活动轮廓模型,由于其良好的特性,在图像分割领域已经有了广泛的应用。本文对于一种耦合的活动轮廓模型,应用变分法求解出其对应的水平集曲线演化的偏微分方程,并给出了解的存在惟一性证明,然后将它应用到图像分割和目标轮廓提取中,取得了良好的效果。On the basis of level set methods, curve evolution and active contour models, geometric active contour models come into being. It has been widely used in image segmentation for its good characteristics. Toward a given coupled active contour models, this paper seeks the partial differential equations corresponding to the level set evolution according to variational principle and gives the proof of the existence, uniqueness of the solution. At last, it has been applied to image segmentation and boundary extraction. The results show that the new approach is more accurate and robust and can obtain very good effect.

关 键 词：图像分割 变分方法 水平集 活动轮廓模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]