一种基于Canny边缘的红外与可见光图像配准算法  

Canny edge based registration algorithm of IR and visible images

作　　者：周锋飞[1] 陈卫东[1] 李良福[1] 

机构地区：[1]西安应用光学研究所,陕西西安710065

出　　处：《应用光学》2009年第4期605-609,共5页Journal of Applied Optics

摘　　要：针对异源传感器在图像配准时如何精确地找到相应的匹配特征问题,结合红外图像与可见光图像的特征,利用2种传感器对同一场景成像时具有一定的边缘轮廓相似特性,提出了提取图像的Canny边缘,在边缘轮廓上根据边缘曲线上像素点的位置夹角的相似性寻找对应匹配点,通过仿真和实验数据与真实值的比较验证算法的有效性和配准的精度。Since the problem of how to accurately find the corresponding matching characteristics by the heterologenetic sensors when image registration, in combination with the characteristics of the red and visible images, the extraction of the Canny edge of images is proposed by utilizing the similar characteristics of the edge outline when the different two sensors image a same scene. That is, find the corresponding matching point at the edge outline according to the similarity of the location included angle between the pixels on the edge curves. The validity and registration accuracy of the algorithm were verified through the simulation and comparison of the experimental data with the true value.

关 键 词：图像配准 Canny边缘 HAUSDORFF距离 特征匹配 

分 类 号：TN219[电子电信—物理电子学]