基于新的加权互相关的图像匹配  

Image Matching Based on New Weighted Normalized Cross Correlation

作　　者：杨兵[1] 于秋则[1] 刘永才[1] 田金文[1] 

机构地区：[1]华中科技大学多谱信息处理技术国防科技重点实验室,武汉430074

出　　处：《弹箭与制导学报》2008年第4期199-202,共4页Journal of Projectiles,Rockets,Missiles and Guidance

摘　　要：为了提高红外图像和可见光图像匹配的鲁棒性,提出一种新的基于结构张量的加权互相关算法。传统的基于边缘的归一化互相关算法仅仅利用边缘点进行匹配,新算法不仅利用边缘点,而且将图像边缘点附近的结构信息嵌入到匹配算法中。实验结果表明,新算法在抗噪能力、匹配性能等方面均优于传统的归一化互相关算法。In order to improve the robustness of image matching between infrared images and visible images, a novel weighted cross correlation algorithm based on structure tensor was presented. Unlike traditional normalized cross correlation algorithm which only use edge points for matching, our new method imbedded local image information into cross correlation algorithm. Experimental results show that the proposed algorithm outperforms ordinary normalized cross correlation algorithm not only in antinoise but also in matching capacity.

关 键 词：归一化互相关 结构张量 图像匹配 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]