基于MRI序列图像的分割方法及三维网格剖分模型的构建.pdf

河北工业大学硕士学位论文


基于 MRI 序列图像的分割方法及三维网格剖分模型的构建

摘要

医学图像分割和三维重建是医学图像研究领域的重要内容。本文工作的目的是探讨基
于 MRI 序列图像的分割方法和三维网格模型的重建。分析了当前图像分割以及三维重建的
各种方法和研究现状,并在此基础上综合运用阈值分割以及数学形态学算法实现对脑组织
磁共振图像序列的分割。首先对图像的预处理分析了邻域平均法、中值滤波方法和维纳滤
波方法,具有统计意义的维纳滤波对图像处理结果较好。对阈值选取采用迭代阈值、最大
类间方差法、二维最大熵求取阈值三种方法,并对其进行了比较,同时运用数学形态学算
法对 MRI 序列图像进行分割。实验证明该方法简单有效,运算速度快,同时能够满足感兴
趣区的分割要求。在图像三维重建方面,着重研究了切片级重建,对于给定的 MRI 医学序
列图像,通过轮廓跟踪提取边界点,再有效结合曲率法和等间距采样法提取特征点,将 MRI
脑图像的各层轮廓用最短对角线法约束条件来进行三角面片的拼接,实现了 MRI 序列图像
的网格模型的建立。最后改进移动立方体算法的跟踪速度,重建了头和脑的真实感模型。

关键词:阈值分割,形态学算法,表面绘制,三维重建
i

基于 MRI 序列图像的分割方法和三维网格剖分模型的构建


IMAGE SEGMENTATION ALGORITHM AND 3D MESH GENERATION MODEL
RECONSTRUCTION BASED ON MRI IMAGE SEQUENCE

ABSTRACT

Medical image segmentation and 3D reconstruction are important in medical image research
field. The major goal of this dissertation is to explore an algorithm for medical image
segmentation and 3D mesh generation model reconstruction based on MRI images of the brain.
We analyzed the backgound and status of above-mentioned algorithms, and based on this, we
comprehensively used the threshold segmentation algorithm and mathematical morphology
algorithm to realize the segmentation of the brain ic resonance image sequence. On the
threshold selection ,pared iterative threshold, the largest category of variance and
two-dimensional maximum entropy to seek the best threshold. At the same time we used
mathematical morphology algorithm to handle the sequence of the MRI images . Experiment
proved that the method is high speed, simple and effective, and it can meet the requirements of
separate to the interest areas. It can obtain the satisfactive segmentation result to the humanity vision
system characteristic .In the three-dimensional image reconstruction, we focused on a section-level
reconstruction. For a given MRI medical image sequences, through the contour track we can
obtain the border of the image, and then bin