基于自适应局部CV模型的图像分割  

Image Segmentation Based on Adaptive Local CV Model

作　　者：马春梅[1] 郑倩[2] 卢振泰[2] 

机构地区：[1]曲阜师范大学计算机科学学院 [2]南方医科大学生物医学工程学院

出　　处：《电子技术（上海）》2012年第11期25-27,共3页Electronic Technology

基　　金：国家自然科学基金(31000450)

摘　　要：文章提出了一种自适应的局部CV模型。局部CV模型克服了传统CV模型处理非均匀图像中不能准确分割的缺点,尽管模型中引入了局部邻域信息,但是也附带分割结果对邻域大小的敏感性和手动设置合适的窗口,文章利用局部方差自动判定局部邻域的窗口大小,解决了分割结果对邻域大小的敏感性,同时避免了手动调节合适的窗口大小。对模拟和真实图像分割的实验结果证明,此方法对于非均匀图像可以给出较为准确的分割结果。An adaptive local CV model is proposed in this paper. The presented method can segment the medical image which has intensity inhomogeneity, but the incorporation of neighborhood information in the adaptive CV model take the difficult of selecting the proper size of neighborhood information simultaneously. Instead of selecting a fixed size, we use the variance of the local region to select the window size of local neighborhood automatically, reduce the sensitivity to the size of neighborhood information, and avoid the manual adjustment of suitable window size. The experimental results on the segmentation of simulated and real images indicate that the proposed method can offer more accurate segmentation result for the inhomogeneous images.

关 键 词：CV模型 方差 分割 邻域信息 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]