摘要： 针对三维图像多相分割问题，提出一种变分水平集分割方法。由变分方法和梯度坡降方法得到能量泛函取极小值的水平集函数演化方程，与基于区域模型的参数估计构成一个交替迭代过程。仿真结果表明，该方法简单高效，能快速实现三维图像的轮廓分割与重建，真实反映采集序列断层图像的信息，具有较好的应用价值。

关键词: 图像分割, 水平集分割, 曲面演化方程, 参数估计模型

Abstract: Aiming at the multiphase segmentation problem of 3D image, this paper proposes a variation level set segmentation method. It gets level set function evolution equation when energy functional gains minimal value by variation method and gradient decent method, and composes a alternative implicit process with parameter estimation based on locale model. Simulation results show that this method is simple and efficient, it can quickly realize outline segmentation and reconstruction of 3D image and truly information reflecting collecting sequence fault image.

Key words: image segmentation, level set segmentation, planar evolution equation, parameter estimation model

中图分类号: 

• TP391.41