基于互信息的图像配准的设计与实现.doc

基于互信息的图像配准的设计与实现 院 系 专 业 班 级 学 号 姓 名 指导教师 负责教师 摘 要 ( 互信息) , 随着两幅图像对齐程度的变化，这种关联也随之变化。当互信息达到最大时，则认为两幅图像已配准，互信息法已经成为图像配准的事实标准。再利用互信息和Powell 优化算法实现了多模态图像的配准，并且达到了亚像素精度。 二、基于互信息的配准方法直接利用图像的灰度值实现两幅图像间的配准。 三、利用相似性测度是用来度量参考图像和待配准图像中提取的两个特征集之间的相似性，能够衡量每次变换结果优劣的准则，本文应用相似性度量为互信息理论。 四、利用一维搜索就是求目标函数在直线上的极小值点，也称线性搜索。一维搜索是许多非线性搜索的重要组成部分，是研究的Powell算法的基础。 五、确定搜索空间和搜索策略。搜索空间与相似性度量密切相关，不同的搜索空间对应不同的相似性度量。搜索策略的选择直接关系到配准速度的快慢。 关键词：Mutual Information Based Image Registration Of The Design And Implementation Abstract Image registration is often as other image processing applications, the use of pre-treatment step is often used for image alignment, target identification and location. Image registration is a variety of image processing and application of foundation, registration results will directly affect their subsequent image processing results. The mutual information registration algorithm based on the principle. The use of mutual information for image registration has become a hot area of ??image processing. Image fusion is the key to image registration. Image registration methods are feature-point method, curve, surface normal and the moment axis method. In this process does not require image segmentation and feature extraction, image registration can be achieved automation, and robustness of a strong, high registration accuracy, but the calculation of mutual information similarity is based on the entire image pixel gray scale, Therefore, a higher computational complexity. Maximum mutual information registration method by eliminating the need for different imaging modes of the relationship between gray-scale images to make any assumptions, nor the need for image segmentation, or any pre-processing, with a high degree of automation features. In recent years has been more and more attention of scholars, and in the field of image registration has been highly valued and widely used by many image processing software package as a standard alignment algorithm. This ar