图像分割算法理论研究
图像分割算法理论研究(任务书,开题报告,外文翻译,论文10000字)
摘 要
图像分割是计算机视觉领域的基础课题,从上世纪七十年代开始就开始被人们运用在医疗、工业、军事等领域中。计算机视觉的任务是理解图像、识别图像中的目标,图像分割是其中底层的一环。图像分割的目标是把图像分成多个含有相同特征的区域,使图像更便于理解。图像分割是图像识别以及特征提取的前提步骤。
边缘是图像最基本的特征,是现实图像中的不规则结构与不平稳现象,也就是信号的突变处,携带了图像中的大量信息,例如物体边界与形状,阴影纹理这些信息都有边缘产生。我们可以通过检测边缘得到图像中这些部分的轮廓。
图像分割的方法有基于边缘的分割方法,这种方法通常是指基于检测灰度值的突变,经典边缘检测算子有:Sobel,Prewitt,Roberts、Laplace等,通过构造检测算子对图像做卷积运算,根据一阶微分或者二阶微分得到梯度最大值(Sobel等梯度算子)或者二阶导数的过零点(Laplace算子),最后选取阈值得到边界。以及基于特定理论的边缘检测方法,包含小波变换和模糊聚类,小波变换可以对信号进行多分辨率分析,对于非平稳信号很有用。本文对上述这些算法进行了研究,通过对这些算法的实验,总结出了这些算法的适用情况和优缺点。 [资料来源:Doc163.com]
关键词:图像分割;边缘检测;边缘检测算子;小波变换
ABSTRACT
Image segmentation is a fundamental topic in the field of computer vision.It has been used in medical, industrial, military and other fields since the 1970s. The task of computer vision is to understand the image, identify the target in the image, and image segmentation is a part of the bottom layer. The goal of image segmentation is to divide the image into multiple regions with the same features to make the image easier to understand. Image segmentation is a prerequisite step for image recognition and feature extraction.
The edge is the most basic feature of the image. It is the irregular structure and unstable phenomenon in the real image, that is, the sudden change of the signal, carrying a lot of information in the image, such as the boundary and shape of the object, the information of the shadow texture has edge generation. . We can get the outline of different parts of the image by detecting the edges.
The method of image segmentation has edge-based segmentation method. The edge-based segmentation method usually refers to the mutation based on the detected gray value. The classical edge detection operators are: Sobel, Prewitt, Roberts, Laplace, etc. Convolution operation is performed to obtain the gradient maximum value (Sobel equal-gradient operator) or the zero-order derivative of the second-order derivative (Laplace operator) according to the first-order differential or second-order differential, and finally select the threshold to obtain the boundary. And edge detection methods based on specific theory, including wavelet transform and fuzzy clustering. Wavelet transform can perform multi-resolution analysis on signals, which is useful for non-stationary signals. In this paper, the above algorithms are studied. Through the experiments of these algorithms, the application, advantages and disadvantages of these algorithms are summarized.
[资料来源:Doc163.com]
KEY WORDS:Image segmentation; edge detection; edge detection operator; wavelet transform
目 录
目 录 III
第1章 绪论 1
1.1 研究背景和意义 1
1.1.1 数字图像处理 1
1.1.2 图像分割 1
1.2 研究发展现状 2
1.3 本文的研究目标和内容 2
1.4 论文结构 3
第2章 基于边缘检测的图像分割算法 4
2.1 经典算子 4
2.1.1 一阶算子 4
2.1.1.1 Roberts算子 5
2.1.1.2 Sobel算子 6
2.1.1 二阶算子 6
2.2 最优算子 7
2.2.1 Marr和Hildreth边缘检测器 7
2.2.2 Canny检测器 9
第3章 模糊聚类分割法 11
3.1 模糊理论 11
3.2 聚类分析 11
3.3 模糊C均值聚类算法(FCM) 11
3.3.1 模糊C均值聚类算法 11
3.3.2 模糊C均值聚类图像分割算法 12
第4章 基于小波变换多分辨率分析的分割算法 14
4.1小波变换的提出背景 14
4.2 小波变换 14
4.2.1 连续小波变换 15
4.2.2 离散小波变换 15
4.3 多分辨率分析和 Mallat 算法 16
4.3.1 多分辨率分析 16
4.3.2 Mallat算法 17
4.3.3 小波变换模极大值边缘检测法 18
第5章 基于边缘检测算法与基于小波变换的多尺度检测算法的效果比较 20
5.1 边缘检测的评价 20
.5.1.1图像重建方法 20
.5.1.1重建图像与原图的相似度 21
5.2 实验对比 21 [资料来源:http://Doc163.com]
第6章 结束语 24
6.1 论文工作总结 24
6.2 问题和展望 24
致 谢 25
参考文献 26
[版权所有:http://DOC163.com]