基于贝叶斯分类模型的双水平集算法的大脑图像分割

针对大脑图像中灰质和白质边界结构的复杂性以及拓扑细长部分目标和弱边界目标分割存在的问题,提出了基于贝叶斯分类模型的双水平集分割算法.鉴于传统的水平集有分割过度、泄漏边界的缺点,可通过贝叶斯分类模型计算出水平集曲线位于边界的概率,并将此概率相关联的区域决策因子添加在水平集函数方程中,从而实现利用图像的区域信息提高水平集曲线识别边界能力的目的.将基于贝叶斯分类模型的双水平集算法应用到大脑图像的分割,通过内外两条水平集共同演化作用,得到了比贝叶斯分类模型的单水平集方法更完整的分割效果,并明显提高了分割效率.

BRAIN IMAGE SEGMENTATION OF DUAL LEVEL SET ALGORITHM BASED ON BAYESIAN CLASSIFICATION MODEL

In light of the complexity of boundary structure of gray and white matters in brain image and for solving the problem of segmenting the topological slender object and weak boundary object,a dual level set segmentation algorithm based on Bayesian classification model has been proposed.In view of the defects in conventional level set methods such as over-segmenting and leaking boundary,the probability of level set curve lies on boundary could be calculated by using Bayesian classification model,and the region...