摘要
水平集方法在图像分割和计算机视觉领域有很广泛的应用,在传统的水平集方法中,水平集函数需要保持符号距离函数.现有的活动轮廓模型、GAC模型、M-S模型、C-V模型等在演化过程中均需要对水平集函数进行重新初始化,使其保持符号距离函数,然而这样会引起数值计算的错误,最终破坏演化的稳定性,另外这些模型只适用于灰度值较为均匀的图像,对灰度值不均匀的图像不能进行理想的分割·针对这些问题,结合C-V模型的思想,提出了一种带有正则项的四相水平集分割模型,其中正则项被定义为一个势函数,具有向前向后扩散的作用,使水平集函数在演化过程中保持为符号距离函数,避免了水平集函数重新初始化的过程.最后对该模型进行数值实现,实验表明了新模型的可行性和有效性.
The level set method has been widely used in image processing and computer vision.In conventional level set formulations,the level set functions need to keep signed distance function.For active contours,GAC model,M-S model and C-V model,we need to reinitialize level set function as a signed distance function.The reinitialization may cause numerical errors and eventually destroy the stability of the evolution.Besides,the most widely used image segmentation algorithms are rely on the homogeneity of the image intensities in the regions of interest,which often fail to provide accurate segmentation results due to the intensity inhomogeneity.To solve these problems,a four-region level set method with regularization term is proposed due to the idea of C-V model＆#39;s idea.The distance regularization term is defined with a potential function,so that the derived level set evolution has a unique forward-and-backward diffusion effect,which is able to maintain a signed distance function during the evolution of level set and eliminate the need for reinitialization.Finally,we show numerical examples and verify the feasibility and effectiveness of the new model.
出处
《数学的实践与认识》
北大核心
2017年第22期147-153,共7页
Mathematics in Practice and Theory
关键词
水平集方法
四相水平集图像分割模型
正则项
数值实现
Level set
four-region level set image segmentation model
regularization term
numerical examples
