结合局部与全局信息的主动轮廓模型  

Active contour model by combining with local and global information

作　　者：代双语[1,2] 王智峰[1] 张学东[1] 

机构地区：[1]辽宁科技大学电子与信息工程学院,辽宁鞍山114051 [2]锡林郭勒职业学院机械与电力工程系,内蒙古锡林浩特026000

出　　处：《辽宁科技大学学报》2013年第2期145-150,165,共7页Journal of University of Science and Technology Liaoning

基　　金：辽宁省教育厅一般项目(L2011042)

摘　　要：针对基于主动轮廓模型的图像分割,现有算法不能满足其实际应用多样性的问题,结合LBF模型和C-V模型的优点,提出一个新的结合局部与全局信息的LGIF主动轮廓模型,解决了LBF模型容易陷入局部极小值而导致错误分割的问题,也解决了C-V模型演化速度慢的问题。实验证明,新模型能分割C-V模型、CVI模型不能分割的灰度不均匀的图像,也能分割LBF模型不能分割的灰度不均匀的图像,对初始轮廓的大小和位置不敏感,具有较强的抗噪性。新模型演化速度快,耗时短,效率高,稳定性好。For the image segmentation with active contour model,the existing algorithms cannot satisfy the diversity of its practical application.By combining with the advantages of Local Binary Fitting（LBF） model and Chan-Vese（C-V） model,a new model called LGIF model is proposed.The local and global information are used effectively for image segmentation.The new model can solve the problem of LBF model easily falling into local minimum and error segmentation,and the problem of slow evolution speed of C-V model is also solved.The Experiments show that this model can be used to segment the images with intensity inhomogeneous while C-V model and LBF model can not,and LGIF model is not sensitive to the initial contour with strong noise immunity.The new model has the advantages of faster evolution speed,short time,high efficiency and good stability.

关 键 词：图像分割 水平集方法 主动轮廓模型 C-V模型 LGIF模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]