摘要
为了克服局部图像拟合模型对轮廓初始化敏感的不足,结合改进C-V模型,提出一种融合局部和全局图像信息的活动轮廓模型.首先由改进C-V模型的全局灰度拟合力和局部图像拟合模型的局部灰度拟合力的一个线性组合来构造水平集演化力,然后通过调整这2个拟合力的权重以提升该模型对轮廓初始化的灵活性,最后利用高斯滤波正则水平集函数法实现水平集函数的正则化.实验结果表明,对于一些真实和人造图像,文中模型显示了对轮廓初始化的鲁棒性,以及较好地处理灰度不均图像的能力.
An active contour model combining both local and global image information is proposed in this paper to alleviate the high sensitivity on the contour initialization of the local image fitting（LIF） model.First,the force of level set evolution is defined as a linear combination of the global intensity fitting force based on the improved C-V model and the local intensity fitting force based on the LIF model.Then,by appropriately choosing the weights of the forces,our proposed model allows flexible initialization of the contours.Finally,Gaussian filtering regularized level set method is employed to regularize the level set function.Experimental results on both real and synthetic images show that the proposed model is robust to the contour initialization,while having the ability of handling intensity inhomogeneity.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2012年第3期364-371,共8页
Journal of Computer-Aided Design & Computer Graphics
基金
重庆市科委自然科学基金计划资助项目(2010BB9218)
关键词
图像分割
活动轮廓模型
C-V模型
局部图像拟合模型
灰度不均
image segmentation
active contour model
C-V model
local image fitting model
intensity inhomogeneity
