毕业论文(图像匹配).doc

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。对位移量比较大的图像，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。该方法的效果要好，成功率有所提高。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，速度快图像，，，Abstract Digital image registration is the same scene shot from the overlapping region of the image 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 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 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 image 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 up. 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