《毕业论文(图像匹配)》由会员分享，可在线阅读，更多相关《毕业论文(图像匹配)（31页珍藏版）》请在金锄头文库上搜索。
1、 毕业设计（论文）I摘要摘要数字图像配准是指将从同一场景拍摄的具有重叠区域的图像通过特征匹配方法，找出图像之间的对应关系。目前，图像配准技术广泛应用于医学、生物、信息处理和其它领域，它已成为图像处理应用中不可或缺的技术。本文主要主要介绍了灰度相关的配准方法，灰度相关的图像配准算法是图像配准算法中比较经典的一种，很多配准技术都以它为基础进行延伸和扩展。它是从待拼接图像的灰度值出发，对待配准图像中一块区域与参考图像中的相同尺寸的区域使用最小二乘法或者其它数学方法计算其灰度值的差异，对此差异比较后来判断待拼接图像重叠区域的相似程度，由此得到待拼接图像重叠区域的范围和位置，从而实现图像配准。对位移量比
2、较大的图像，可以先校正图像的旋转，然后建立两幅图像之间的映射关系。当以两块区域像素点灰度值的差别作为判别标准时，最简单的一种方法是直接把各点灰度的差值累计起来。另一种方法是计算两块区域的对应像素点灰度值的相关系数，相关系数越大，则两块图像的匹配程度越高。该方法的配准效果要更好，配准成功率有所提高。基于灰度相关的配准方法的优点是原理简单，直观性好，计算量较小；缺点是只适用于具有水平或者垂直位移方向上的两幅图像的配准，根据所选择的模板的不同，基于灰度相关的配准方法可以分为：线模板匹配法、比值模板匹配法和块模板匹配法。本文分别对 3 种匹配方法的算法原理和具体实现进行深入研究。通过研究发现，线模板匹
3、配法精确直观，但是它只能处理简单的平移变换下的图像配准，容易受光照的影响，不能实现图像旋转和缩放情况下的配准。比值模板匹配法算法清晰简单，容易理解，实现起来比较方便。在匹配计算的时候，计算量小，速度快；但是，不能处理旋转缩放情况下的配准。块模板匹配法能准确找到匹配点，能在小的旋转和缩放变形的情况下实现配准；但计算量较大。关键词关键词图像配准，灰度相关，线模板匹配法，比值模板匹配，块模板匹配毕业设计（论文）IIAbstractDigital image registration is the same scene shot from the overlapping region of the i
4、mage by a feature matching method to find the corresponding relationship between the images. Currently, image registration techniques are widely used in medicine, biology, information processing and other area. It has become an integral part of image processing applications of technology. This paper
5、 introduces registration method based gray-correlation. The mothod is more classic one, many registration techniques are were extended and expanded based on it. It is a mosaic image from the gray value to be proceeding towards registration with the reference image in a region of the same size in the
6、 image area using the least squares or other mathematical methods to calculate the gray value differences, this comparison was to determine differences overlap the image mosaic to be the degree of similarity, the resultant mosaic image to be the extent and location of overlap, in order to achieve im
7、age registration. Larger than the image of the displacement can be corrected image rotation, then establish the mapping between two images. When two regional differences in gray value of pixels as the criterion, the most simple way is to point directly to the cumulative difference between the gray u
8、p. Another method is to calculate the two regions of gray value pixel corresponding to the correlation coefficient, correlation coefficient is larger, the higher the degree of matching two images. Effect of the method by better registration, registration has increased the success rate. Based on gray
9、 correlation registration method has the advantage of simple in principle, straightforward, good, less computation; disadvantage is only applicable to the direction of a horizontal or vertical displacement of two images on the registration, according to the different choice of templates based on gra
10、y correlation registration method can be divided into: line template matching, template matching method and the block ratio template matching method. This paper on the line template matching, template matching method and the block ratio template matching algorithm principle and concrete achieve-dept
11、h study. It has found that line template matching method has precise visual, but it can handle only simple translation transformation of the image registration is volunerable can not be achieved in case of registration. Template matching algorithm is the ratio of clear and simple, more convenient to
12、 implement. When the matching calculation, computation speed; however, can not handle the rotation scaling in case of registration. Block template matching method can accurately find the matching points that can rotate and zoom in the small deformation registration under the situation.Key wordsImage
13、 registration, intensity-related, line template matching method, the ratio of 毕业设计（论文）IIItemplate matching, block matching, rotating around the optical axis毕业设计（论文）IV目录第第 1 1 章章 绪论绪论.1 11.1研究背景及意义.11.2图像配准方法概述.21.3研究现状.31.4研究问题及内容.4第第 2 2 章章 图像配准基本理论图像配准基本理论.1 12.1图像配准的基本介绍.12.2图像配准的相关概念.12.3灰度相关的配准方法.5第第 3 3 章章 线匹配法线匹配法.7 73.1线匹配法基本介绍及原理.
