基于相关系数的快速图像匹配研究  

Fast Image Matching Based on Correlation Coefficient

作　　者：李卓[1] 邱慧娟[1] 

机构地区：[1]北京理工大学信息科学技术学院光电工程系,北京100081

出　　处：《北京理工大学学报》2007年第11期998-1000,共3页Transactions of Beijing Institute of Technology

基　　金：国家部委预研项目(BQ0185)

摘　　要：针对传统图像相关匹配算法计算量较大的问题,研究图像库中相关图像搜索方法.提出一种改进的相关匹配算法,该方法通过对相关系数公式进行简化和迭代运算,减少了重复运算.通过设定参考图像与目标图像的相关系数阈值,只需计算方差相差较小的点的相关系数,提高了运算速度,计算时间减少到原来的14%,甚至更短.在文件夹的图像搜索中实现了图像的快速匹配.As the traditional method of image matching algorithm involves a large amount of calculation, a modified correlation matching algorithm is proposed. The correlation coefficient is simplified, using iterative calculation and reduced duplication in operation. By setting the correlation coefficient threshold, only the point is calculated where the difference of variance between the correlation coefficient point is small, thus the speed is increased. Computing time can be reduced to only 14 % of the original or less. In the folder of image search are realized fast matching of images.

关 键 词：灰度相关 图像匹配 方差 迭代算法 像素抽样 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]