中国科学院沈阳自动化研究所机构知识库
Advanced  
SIA OpenIR  > 工业信息学研究室  > 先进制造技术研究室  > 学位论文
题名: 医学图像分割及二维轮廓数据处理技术研究
其他题名: Reseach on Medical Image Segmentation and relevant Data Processing
作者: 张慧
导师: 刘伟军
分类号: TP391.4
关键词: 图像分割 ; 水平集 ; 形状先验知识 ; 小波
索取号: TP391.4/Z32/2005
学位专业: 模式识别与智能系统
学位类别: 硕士
答辩日期: 2005-05-31
授予单位: 中国科学院沈阳自动化研究所
学位授予地点: 中国科学院沈阳自动化研究所
作者部门: 先进制造技术研究室
中文摘要: 图像分割是三维表面重建的一个基本步骤,也是从原始图像数据到三维实体过渡的基础。传统的分割方法由于其自身的局限性难以达到理想的分割效果,对后续的重建带来很大误差。尤其是医学图像,含有复杂的人体组织结构,本质上具有灰度上的含糊性和不均匀的特点,再加上成像设备质量问题产生的噪声,常规的分割方法很难满足重建的需要。基于主动轮廓的图像分割方法是近年来提出的结合图像信息和模型自身特性的一种高层次的分割方法,经典的 snake分割方法就属于早期的主动轮廓分割方法,但snake方法因无法分割拓扑形状改变的目标而使应用受到了限制。水平集正是在这种情况下被引入主动轮廓分割领域并得到广泛研究。水平集方法是处里封闭运动界面随时间演化过程中拓扑结构变化的有效工具,它克服了传统分割方法无法改变拓扑形状的弊病。 Mumford-shah模型是解决图像中自由不连续问题的优秀模型,在图像除噪、恢复、形状匹配尤其是分割领域得到了极大的应用,将水平集方法和Mumford-shah模型结合对分割复杂图像中的目标具有天然的优势,它的优势就在于不依赖于图像的梯度信息,对边缘模糊、不连续的目标能够得到比较好的分割效果。尽管将水平集与Mumfor-shah模型相结合的分割方法克服了拓扑形变及依赖图像边缘的缺陷,但仍然无法排除具有相似灰度的其它组织的干扰。也就是说,仅利用图像本身的信息很难满足精确分割医学图像中目标的要求,因此有必要加入目标的形状先验知识指导分割。将目标的形状先验知识引入水平集分割是一个非常崭新的领域,围绕两个基本问题:1.如何描述目标的形状;2.如何将形状先验知识加入到曲面演化过程中指导水平集分割;人们提出了不同的分割模型。分割后得到的目标边缘轮廓数据量非常大,除含有特征信息外,还含有大量的冗余数据,如不精简,将给后续的三维表面重建带来很大麻烦。关于轮廓数据的精简,有很多种算法,常用的如均匀抽样法,DP算法等,但这些算法在保存原始轮廓数据的特征信息方面不够理想。本文以水平集的曲线演化模型为基础,首先探讨了水平集的数值解和迭代公式,提高水平集演化速度的窄带算法、快速步进法;在介绍了Mumford-shah模型的理论基础上,详细阐述了无边缘检测算子的水平集分割方法,及其数值计算公式;介绍了当前国外几种流行的与水平集结合的统计形状模型,围绕形状模型的建立和形状模型如何在水平集的分割过程中起作用进行了讨论,同时提出了一种新的基于统计形状模型的水平集分割算法。为满足后续三维重建的需要,本文通过对分割后得到的轮廓特征点的分析,提出了一种基于小波变换的特征点提取和精简算法,该算法能够排除噪声的干扰精确的定位到特征点,同时能够达到数据精简的目的。文章给出了算法的步骤。为验证本文所提算法的可行性与有效性,在实验中选择了脊柱骨和下颚骨的CT图像进行分割,这两组图像存在拓扑形变,边缘模糊等特点,实验结果对比了的各种算法的优缺点,显示了本人提出算法的优越性。
英文摘要: Image segmentation is a basic problem converting the original image to 3D solid surface representation. It is also a pre-step to accomplish the 3D surface reconstruction. The traditional segmentation method, however, can’t got ideal result due to their own limitations, especially in medical image segmentation area, the image contains complex tissues and structures, and the image itself are inevitably deteriorated by the noise due to the imaging equipment. So, the classical segmentation method can’t satisfy our need. The active contour method is a high level segmentation method combining the image information and the model feature. The classical snake model are difficult to handle the change of topology structure, which limit it’s biomedical application. The level-set method based on geometric contour model is thus came into use and got extensive study thanks to it’s properties of topology adaptable. Level-set method is initialed from the propagation of the front as a useful computing tools to deal with the propagation of closed areas, which has many strong point in segmentation areas. Mumford-shah model is an excellent model in dealing with the un-continuous problems of image which has already got extensive use in image de-noising, recovering, registration and especially in segmentation. The combination of Mumford-shah and level-set method has many advantages to segment complex object since this method is independent of the gradient of the image. it can also detect object with break or blur boundary. In order to segment a particular kind of object for subsequent reconstruction, We must get rid of noise and the disturbance of object with similar grey, it demand the incorporation of as much as prior information as possible to help the segmentation algorithms extract the tissues of interest. Incorporating prior information to level set segmentation is a new area, based on the two main problem: how to represent the shape; how to incorporate shape into the propagation procedure. Different researchers have establish different models. A new wavelet based algorithm is proposed to extract feature point and decrease redundant point in order to make preparation for the subsequent 3D surface reconstruction. All the method we proposed is based on level-set theory, including it’s numerical method, narrow banding method, the fast marching method, we also give the method of active contour without edges based on Mumford-shah model and it’s numerical iterative steps. After introducing several shape prior model based on level-set, we proposed our new method for the level-set segmentation. We select two set of spine and chin images to testify the methods introduced in this article and the algorithm we proposed in chapter5, we then give experiment result to compare different kind of methods, which also shows the feasibility and validity of our algorithm.
语种: 中文
产权排序: 1
内容类型: 学位论文
URI标识: http://ir.sia.cn/handle/173321/9578
Appears in Collections:工业信息学研究室_先进制造技术研究室_学位论文

Files in This Item:
File Name/ File Size Content Type Version Access License
医学图像分割及二维轮廓数据处理技术研究.pdf(2199KB)----限制开放 联系获取全文

Recommended Citation:
张慧.医学图像分割及二维轮廓数据处理技术研究.[硕士学位论文].中国科学院沈阳自动化研究所.2005
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[张慧]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[张慧]‘s Articles
Related Copyright Policies
Null
Social Bookmarking
Add to CiteULike Add to Connotea Add to Del.icio.us Add to Digg Add to Reddit
所有评论 (0)
暂无评论
 
评注功能仅针对注册用户开放,请您登录
您对该条目有什么异议,请填写以下表单,管理员会尽快联系您。
内 容:
Email:  *
单位:
验证码:   刷新
您在IR的使用过程中有什么好的想法或者建议可以反馈给我们。
标 题:
 *
内 容:
Email:  *
验证码:   刷新

Items in IR are protected by copyright, with all rights reserved, unless otherwise indicated.

 

 

Valid XHTML 1.0!
Copyright © 2007-2016  中国科学院沈阳自动化研究所 - Feedback
Powered by CSpace