基于高斯混合模型的医学图像分割技术研究.doc

基于高斯混合模型的医学图像分割技术研究

1.55万字
自己原创的毕业论文，已经通过校内系统检测，重复率低，仅在本站独家出售，大家放心下载使用

摘要 医学图像分割算法的研究是当前医学图像处理和分析的热点，医学图像分割主要是指将图像分成各具特性的区域并提取出感兴趣目标的技术。由于医学应用对医学图像分割的准确度和分类算法的速度要求较高，而人体解剖的个体差异较大，一般来说由于图像本身受到噪声、局部体效应以及偏移场效应等的影响，往往使得一般意义上的图像分割算法并不能达到理想的效果。其中高斯混合模型越来越受到人们的关注，已经慢慢成为是具有代表性的一种聚类分割方法。期望最大化(Expectation Maximization，EM)算法为模型参数提供了一种简单有效的极大似然迭代求解方法，本论文主要分以下几个方面对基于高斯混合模型的图像分割进行了研究讨论。
1.对国内外医学图像分割方法进行了介绍，介绍了高斯混合模型及EM算法以及EM算法在高斯混合模型的应用，并重点对基于高斯混合模型的图像分割方法进行了分析。阐述并展示了一个具体的数据集，为论文的深入研究提供基础。
2.介绍了MATLAB语言的特点以及如何利用MATLAB及其图像处理工具箱进行图像分割。数字图像的信息量很大，对处理速度的要求也很高，而MATLAB的基本数据单位是矩阵，故常用MATLAB进行图像处理。
3.在迭代的过程中贝叶斯后验概率的干扰，以及医学图像信息本身的复杂性和受到的影响，而EM算法虽然简单容易理解，但仍有不足之处。针对基本的EM算法存在的问题，我们要想办法减少对初始值的依赖和计算出的贝叶斯后验概率对模型分量间的干扰，故论文引入了FCM算法以解决这个问题。实验结果表明，该方法对算法收敛颇有效果，在大部分情况下都得到了较好的分割精度，有效提高图像分割整体性能。

关键词：图像分割 高斯混合模型 EM算法 FCM算法。


Medical Image Segmentation Based on Gaussian Mixture
Model

ABSTRACT Research Summary medical image segmentation algorithm is still current hot medical image processing and analysis , medical image segmentation mainly refers to an image into regions , each with features and technology to extract the object of interest . However, due to the large individual differences in human anatomy , and medical applications to speed medical image segmentation and classification accuracy of the algorithm are higher, and because the image itself is affected by noise, offset and local body effects such as field-effect , making the segmentation algorithm nearly did not achieve the desired scattered fruit . Which is a Gaussian mixture model clustering segmentation Taiwan representative methods , more and more people's attention. Expectation-maximization (Expectation Maximization, EM) algorithm for the model parameters to provide a simple and effective method for solving ML chosen generations , the paper image segmentation Gaussian mixture model-based units were studied .
     1 .for domestic and foreign medical image segmentation methods Dingjie Shao , focusing on the image segmentation method based on Gaussian mixture model -depth analysis . Zhou Shao T Gaussian mixture model and EM algorithm described in the application of the EM algorithm Gaussian mixture model , and shows a real guillotine specific datasets for further research in this paper provides a foundation.
2.introduces the MATLAB language features and how to use MATLAB and Image Processing Toolbox for image segmentation. A large amount of information digital images, processing speed requirements are high, and MATLAB basic data unit is the matrix, it is commonly used MATLAB for image processing.
      3. Interference in the process of iteration Bayesian posterior probabilities, and the complexity and impact of medical image information itself, the EM algorithm is simple and easy to understand, but there are still shortcomings. For the basic problems of the EM algorithm, we find a way to reduce dependence on the initial value and calculate the Bayesian posterior probability of interference between model components, so the paper introduces the FCM algorithm to solve this problem. Experimental results show that the algorithm converges quite effective method, in most cases have been better segmentation accuracy, improve the overall performance of image segmentation.

Keywords : Image segmentation stick Gaussian model EM algorithm algorithm FCM.


目录

摘要 I
ABSTRACT II
第一章 绪论 1
1.1 课题研究背景及意义 1
1.2 国内外研究现状 1
1.3 存在问题 2
1.4 论文的内容安排 3
第二章 高斯混合模型及EM算法 4
2.1 高斯混合模型概述 4
2.1.1 单高斯模型 4
2.1.2 高斯混合模型 4
2.2 EM算法 7
2.2.1 概述 7
2.2.2 算法原理 7
2.2.3 算法性质 8
2.2.4 采用EM估计GMM的参数 9
2.2.5 EM算法在高斯混合模型的应用 10
2.3 本章小结 11
第三章 MATLAB基本知识介绍 12
3.1　MATLAB概述 12
3.2　MATLAB语言的特点 12
3.3　MATLAB在图像处理中的应用 13
3.4 本章小结 13
第四章 高斯混合模型医学图像分割方法 14
4.1 图像的几何模糊性 14
4.2 模糊C均值聚类方法 15
4.4 实验结果 18
4.5 本章小结 20
第五章 总结与展望 21
5.1 总结 21
5.2 展望 21
致谢 23
参考..