基于改进型水平集的医学图像分割研究

发布时间：2018-04-29 16:15

本文选题：医学图像分割 + 水平集　；参考：《扬州大学》2017年硕士论文

【摘要】：图像分割是我们图像处理的关键,是图像处理技术中的底层技术,经过这几十年的发展,在实际生活工作中,图像分割已经得到了广泛的应用。随着医学影像学的发展,医学图像分割也成为了图像分割研究的热点和难点。由于医学图像具有复杂多样性,容易受各种外在和内在因素的干扰产生,分割图会出现噪声、灰度不均、弱边缘、多目标等等,经常需要采用多种方法综合起来对图像进行分割,才能将图像的所有信息充分分割出来。水平集方法(LSM)凭借其算法独特的优势,近年来已成为图像分割领域关注的焦点,并得到了快速的发展和应用。但是,水平集分割方法对处理复杂情况下的图像也会存在一些问题,比如计算量大、鲁棒性、稳定性等,因此开展水平集方法的医学图像分割研究具有很高的应用价值和重要的现实意义。在借鉴诸多文献的基础上,本文进行了基于改进型水平集的医学图像分割研究,并提出了几种改进方法:基于抑制式聚类算法的改进型双水平集图像分割算法、基于偏移场和改进型水平集的图像分割算法等,编写设计了智能图像处理系统。主要研究内容和创新点包括:1、阐述了医学图像分割技术研究的背景和意义,给出了医学图像分割技术的数学描述,介绍了医学图像分割技术研究的现状,同时也讨论了医学图像分割技术的发展和难点。文中用数学方式描述了水平集的基本模型和常见模型。2、医学图像中经常出现噪声、多目标现象,为了能够同时分割多相图像,并对图像噪声干扰具有抑制能力,本文提出基于聚类算法的改进型双水平集模型,首先通过抑制式模糊聚类算法(SRFCM)“抑制式竞争”思想提高算法的收敛速度,然后采用增加能量惩罚项的方式改进双水平集医学图像分割模型(DCV)分割多目标。3、针对医学图像灰度不均,存在偏移场的问题,本文提出基于区域的偏移场耦合模型,通过添加区域信息的方式,将全局分割方法和Split-Bregman方法结合起来。采用N个水平集划分2N个区域变分多相水平集方法,在函数中增加能量惩罚项使水平集避免重新初始化,降低分割算法的计算和时间复杂度。4、编写设计了图像处理智能分割系统V1.0、医学图像处理分割系统V1.0,采用用户引导式、指示型流程,开源性设计方案,程序员可以在MATLAB软件环境Editor中进行算法编译和函数回调,根据实际使用的情况和用户的反馈,进一步改良分割算法和完善用户界面(GUI)。
[Abstract]:Image segmentation is the key of image processing and the bottom technology of image processing. After decades of development, image segmentation has been widely used in real life. With the development of medical imaging, medical image segmentation has become the focus and difficulty of image segmentation. Because of the complexity and diversity of medical image, it is easy to be interfered by various external and internal factors, and the segmentation image will appear noise, uneven grayscale, weak edge, multi-target and so on. In order to segment all the information of an image, it is often necessary to use a variety of methods to segment the image. Level set method (LSM) has become the focus in the field of image segmentation due to its unique advantages in recent years, and has been rapidly developed and applied. However, the level set segmentation method also has some problems in processing images in complex situations, such as large computational complexity, robustness, stability, and so on. Therefore, the research of medical image segmentation based on level set method has high application value and important practical significance. On the basis of many references, this paper studies the medical image segmentation based on the improved level set, and puts forward several improved methods: the improved two-level set image segmentation algorithm based on the suppression clustering algorithm. An intelligent image processing system is designed based on offset field and improved level set image segmentation algorithm. The main research contents and innovations include: 1, the background and significance of medical image segmentation technology are expounded, the mathematical description of medical image segmentation technology is given, and the present situation of medical image segmentation technology is introduced. At the same time, the development and difficulties of medical image segmentation technology are also discussed. In this paper, the basic model and common model. 2 of the level set are described mathematically. In order to segment the polyphase image simultaneously and suppress the noise interference, there is often noise and multi-target phenomenon in the medical image. In this paper, an improved bi-level set model based on clustering algorithm is proposed. Firstly, the convergence speed of the algorithm is improved by the idea of "suppressive competition" of the suppressed fuzzy clustering algorithm (SRFCM). Secondly, the bi-level set medical image segmentation model (DCV) is improved by adding energy penalty term. Aiming at the problem of uneven grayscale and migration field of medical image, a area-based migration field coupling model is proposed in this paper. The global segmentation method and the Split-Bregman method are combined by adding the region information. By using the method of dividing N level sets into 2N regions, the energy penalty term is added to the function to avoid the reinitialization of the level set. Reduce the computation and time complexity of segmentation algorithm. 4. Design the image processing intelligent segmentation system V1.0, medical image processing segmentation system V1.0, using user-led, indicative flow, open source design scheme, The programmer can compile algorithm and callback function in MATLAB software environment Editor, according to the actual use and user feedback, further improve the segmentation algorithm and improve the user interface.
【学位授予单位】：扬州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【相似文献】