大连海事大学
毕业论文Array
二○一三年六月
图像区域分割算法设计
专业班级：2009级自动化(2)班*名：***
指导教师：***
信息科学技术学院
摘要
数字图像处理技术是一个跨学科的领域。

随着计算机科学技术的不断发展，图像处理和分析逐渐形成了自己的科学体系，新的处理方法层出不穷，尽管其发展历史不长，但却引起各方面人士的广泛关注。

首先，视觉是人类最重要的感知手段，图像又是视觉的基础，因此，数字图像成为心理学、生理学、计算机科学等诸多领域内的学者们研究视觉感知的有效工具。

其次，图像处理在军事、遥感、气象等大型应用中有不断增长的需求。

本文介绍了数字图像处理技术中图像分割技术的基本原理和主要方法，对经典的图像分割算法进行了较全面的叙述。

图像分割就是把图像分成若干个特定的、具有独特性质的区域并提出感兴趣目标的技术和过程。

它是由图像处理到图像分析的关键步骤。

现有的图像分割方法主要分以下几类：基于阈值的分割方法、基于区域的分割方法、基于边缘的分割方法以及基于特定理论的分割方法等。

本文根据MATLAB的GUI相关知识建立了显示界面，可以对图片进行预处理、多种方法的分割、显示灰度直方图和保存分割的图片。

全文先后用阈值法、边缘检测法和区域提取法进行了仿真，对分割出来的图片加以比较，探讨各种分割法的优点与不足。

实验证明，没有一种方法是通用的分割理论，因此现已提出的分割算法大都是针对具体问题的，并没有一种适合于所有图像的通用的分割算法。

但是可以看出，图像分割方法正朝着更快速、更精确的方向发展，通过各种新理论和新技术结合将不断取得突破和进展。

关键词：图像分割；阈值；边缘检测；区域生长
ABSTRACT
Digital image processing technology is an interdisciplinary field. With the continuous development of computer science and technology, image processing and analysis, gradually formed its own scientific system.And the new treatment methods emerge in endlessly. Although its history is not long, but cause the extensive concern of all aspects of people. First of all, the visual is the most important human cognitive means, is the basis of the visual image, as a result, digital images become the psychology, physiology, computer science, and many other scholars in the fields of effective tool for the study of visual perception. Secondly, the image processing in the military, remote sensing, meteorology and other large applications with growing demand.
Digital image processing techniques were introduced in this paper with its the basic principle and main methods of image segmentation technology in the classic image segmentation algorithm for the more comprehensive narration. Image segmentation is to divide the image into several specific area, unique quality and interested target technology and process are put forward. It is a key step from image processing to image analysis. Existing mainly in the following categories: threshold segmentation method, region segmentation, edge detection and segmentation method based on specific theory, etc. In this paper, based on MATLAB GUI knowledge display interface is established. It can be for image preprocessing, a variety of methods of segmentation,showing the gray level histogram and saving the segmentation of images. The text has used threshold method, the edge method and region extraction method has carried on the simulation, the partition of the image are compared, the advantages and shortcomings of all kinds of segmentation are discussed.
Experiments show that no one approach is general theory of segmentation, therefore we has proposed segmentation algorithm is mostly directed at specific issues. There is not a general segmentation algorithm that is suitable for all images. But it can be seen that image segmentation method is moving toward more rapid, more accurate direction, through a variety of new theory and new technology will continue to make a breakthrough and progress.
Keywords: Image segmentation; Threshold; Edge detection; Region growing。