摘要
针对乳腺核磁共振成像的灰度不均匀现象,提出一种融合全局和局部信息的水平集图像分割方法(global and local combined C_V,GLCCV)。该方法将图像的局部信息融入基于全局信息的Chan-Vese(C_V)水平集方法;根据局部灰度拟合均值占全局灰度均值的比例,构造自适应平衡指示函数调节全局和局部效应之间的均衡;加入惩罚项以避免重新初始化。对比实验表明,该水平集分割模型能够有效分割多种灰度不均匀场景下的乳腺MR图像,在抗噪和精确性方面优于融合前的分割方法。
To effectively perform breast magnetic resonance imaging（MRI） segmentation in the case of intensity inhomogenei- ty, the paper proposed a novel level set method combined global and local information（GLCCV）. Local information of image was incorporated into the Chan-Vese（C_V） model, which was based on global information. According to the proportion of the local intensity fitting term accounts for the global term, this paper presented a self-adaptive indicator function to balance the global and local effect, added penalty term to avoid re-initialization. Contrast experiments show that the proposed method is ef- ficient to segment breast MRI in variety case of intensity inhomogeneity scenes, and is better than conventional methods in noise resistance and accuracy.
出处
《计算机应用研究》
CSCD
北大核心
2015年第1期307-311,共5页
Application Research of Computers
基金
陕西省科学技术研究发展计划项目(2012K06-36)
陕西师范大学中央高校基本科研业务费资助项目(GK201102006)
关键词
乳腺MRI
融合全局和局部信息
水平集
灰度不均匀
自适应指示函数
breast MRI
combined global and local information
level set method
intensity inhomogeneity
self-adaptive indicator function
