红外与可见光图像配准算法  

Infrared and Visible Image Registration by Combination of Second-order Mutual Information and Alignment

作　　者：匡雅斌[1] 王敬东[1] 李鹏[1] 

机构地区：[1]南京航空航天大学自动化学院,江苏南京210016

出　　处：《电子科技》2011年第5期80-84,共5页Electronic Science and Technology

摘　　要：提出把二阶互信息与对齐度相结合作为相似度函数,通过Hough变换提取两幅图像中一致性较高的直线,计算它们的对齐度,最后针对不同的图像特点,调整对齐度在新相似度函数中所占比例进行配准。实验结果表明,新算法与其他几种测度函数相比有更高的精确性和鲁棒性。In this paper,a registration algorithm that combines second-order mutual information with alignment as a similarity function is proposed.Firstly,the consistency lines are extracted from visible/IR images by using Hough transform.Secondly the alignment metric of consistency lines is calculated.Finally,these different images are registered by adjusting the proportion of alignment metric in the new similarity function.Experiment results demonstrate that the proposed similarity function shows higher accuracy and robustness than the conventional similarity functions.

关 键 词：红外 可见光 图像配准 二阶互信息 对齐度 HOUGH变换 

分 类 号：TN219[电子电信—物理电子学] TP391.41[自动化与计算机技术—计算机应用技术]