搜索
分类号密级国际十进分类号（UDC）基于Matlab的医学图像增强与边缘检测算法的实验研究（题名和副题名）袁丽婷（作者姓名）指导教师姓名邱力军副教授指导教师单位第四军医大学生物医学工程系申请学位级别医学硕士专业名称生物医学工程论文提交日期2008.12答辩日期2007.05论文起止时间2005年8月至2008年12月学位授予单位第四军医大学独独创创性性声声明明秉承学校严谨的学风与优良的科学道德，本人声明所呈交的论文是我个人在导师指导下进行的研究工作及取得的研究成果。尽我所知，除了文中特别加以标注和致谢的地方外，论文中不包含其他人已经发表或撰写过的研究成果，不包含本人或他人已申请学位或其他用途使用过的成果。与我一同工作的同志对本研究所做的任何贡献均已在论文中作了明确的说明并表示了致谢。申请学位论文与资料若有不实之处，本人承担一切相关责任。论文作者签名：日期：保保护护知知识识产产权权声声明明本人完全了解第四军医大学有关保护知识产权的规定，即：研究生在校攻读学位期间论文工作的知识产权单位属第四军医大学。本人保证毕业离校后，发表论文或使用论文工作成果时署名单位仍然为第四军医大学。学校可以公布论文的全部或部分内容（含电子版，保密内容除外），可以采用影印，缩印或其他复制手段保存论文。学校有权允许论文被查阅和借阅，并在校园网上提供论文内容的浏览和下载服务。论文作者签名：导师签名：日期：基于Matlab的医学图像增强与边缘检测算法的实验研究研究生：袁丽婷学科专业：生物医学工程所在单位：第四军医大学生物医学工程系导师：邱力军副教授辅导教师：漆家学高级实验师关键词：图像增强；图像处理；算法；Matlab中国人民解放军第四军医大学2008年12月第四军医大学硕士学位论文目录缩略语表··············································································································1中文摘要··············································································································2英文摘要··············································································································4前言··············································································································7文献回顾··············································································································8正文············································································································20实验一灰度图像的直方图处理·································································201材料·········································································································202方法·········································································································203结果·········································································································234讨论·········································································································28实验二医学图像的滤波处理·····································································291材料·········································································································292方法·········································································································293结果·········································································································314讨论·········································································································33实验三医学图像的边缘检测·····································································341材料·········································································································342方法·········································································································343结果·········································································································384讨论·········································································································40小结············································································································42参考文献············································································································45个人简历和研究成果························································································55致谢············································································································56第四军医大学硕士学位论文-1-缩略语表缩略词英文全称中文全称DSADigitalsubtractionangiography数字减影血管造影MRIMagneticresonanceimaging磁共振成像CRComputedradiography计算机X线摄影DRDigitalradiography数字X线摄影PACSPicturearchivingandcommunicationsystem图像存储和传输系统SPECTSinglephotonemissioncomputedtomography单光子发射体层成像PETPositronemissiontomography正电子发射体层成像DFTDiscretefouriertransform离散傅立叶变换FFTFastfouriertransform快速傅立叶变换DCTDiscretecosinetransform离散余弦变换CWTContinuouswavelettransform连续小波变换DWTDiscretewavelettransform离散小波变换HEHistogramequalization直方图均衡化HRHistogramregulation直方图规定化第四军医大学硕士学位论文-2-基于Matlab的医学图像增强与边缘检测算法的实验研究硕士研究生：袁丽婷导师：邱力军副教授第四军医大学生物医学工程系，西安710032中文摘要目的：本课题以研究图像增强和边缘检测算法的应用为目的，改善医学图像质量，从结果中获取更多有价值的信息，为临床诊断和分析提供依据。医学图像增强主要是为了将图像中感兴趣的特征有选择的突出，具体表现为突出目标物轮廓，增强图像中的细节，提高层次感，滤除各类噪声等等。边缘检测的结果是为了提供图像的重要特性，使医学图像中病灶部位特征能明确区分出来。方法：1、利用Matlab工具箱函数对图像进行预处理，分别采用直方图均衡化和直方图规定化的方法对采集到的医学图像进行处理，然后分析比较处理后的结果图像。2、选择5×5模板和3×3模板，利用Matlab工具箱函数分别对采集到的医学图像用wiener滤波法对图像进行自适应wiener滤波、中值滤波法处理，分析各算法对医学图像的处理效果。3、对采集到的医学图像，结合Matlab具体程序，分别进行Roberts算子、Prewitt算子、Sobel算子、Laplacian算子、Canny算子进行边缘检测，第四军医大学硕士学位论文-3-并分析比较它们的检测结果。结果：1、用直方图均衡化和规定化的算法，将原始图像密集的灰度分布变得比较稀疏，处理后的图像视觉效果得以改善。直方图均衡化对于局部细节增强不显著，而直方图规定化则使不易观察到的细节变得清晰。2、加入高斯噪声后，选择5×5模板的wiener滤波法处理的图像，边缘模糊，图像中软组织的清晰度下降，而选择3×3模板的wiener滤波后的图像视觉效果要好。选择3×3模板的中值滤波后结果图的噪声基本被滤除，图像变的较清晰。3、Roberts算子对边缘定位不准确，Sobel算子和Prewitt算子能够检测出边缘，但边缘是不连续的，且比较模糊，而Sobel算子比Prewitt算子提取细节信息量要多一些，Laplacian算子检测出的边缘产生了虚假信息，Canny算子能真正检测到弱边缘。结论：1、直方图处理医学图像可以有效改善图像质量，使图像变的清晰，直方图规定化处理医学图像局部细节方面好于均衡化。2、wiener滤波法和中值滤波法可以有效地消除医学图像噪声，对医学图像分析和诊断提供了帮助。3、用各种微分算子对医学图像进行边缘提取，分析比较发现Canny算子提取的边缘整个轮廓比较平滑、清晰、完整，噪声抑制较其余微分算子要好。通过以上的工作，本论文讨论的图