基于边缘和互信息的红外与可见光图像配准  

IR/Visible Image Registration Based on Edge Characteristics and Mutual Information

作　　者：孙雅琳[1] 王菲[1] 

机构地区：[1]西安电子科技大学技术物理学院,陕西西安710071

出　　处：《电子科技》2010年第4期80-82,共3页Electronic Science and Technology

摘　　要：采用一种结合边缘特征和互信息的图像配准方法,对红外与可见光图像进行配准。首先,用小波变换提取图像边缘,然后计算两幅边缘图像在不同条件(平移、旋转)下的归一化互信息,取归一化互信息最大时对应的配准参数为所需配准参数,再确定刚性仿射变换模型的参数,最后经过缩放、平移和旋转得到最终配准图像。This paper proposes an image registration method combining edge features with mutual information for the registration infrared and visible image. First, the image edge is extracted by utilizing wavelet. Then the normalized mutual information of two edge images in different conditions, such as translation and rotation, is computed. The registration parameter corresponding to the maximum normalized mutual information is taken as the required registration parameter. Finally, after determining the parameter of the model of rigid affine transformation, the registered image is obtained by scaling, translation and rotation.

关 键 词：图像配准 小波变换 互信息 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]