一种彩色图像配准算法C-GLOH  

A C-GLOH features in color image registration using

作　　者：韩佶轩[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072

出　　处：《电子测量技术》2014年第2期50-53,共4页Electronic Measurement Technology

摘　　要：图像配准是图像三维重构、计算机视觉、模式识别、遥感图像处理等众多领域的关键技术之一。目前将颜色信息运用到图像配准中较为实用的算法是CSIFT,但该算法从速度、效率和鲁棒性上都劣于传统SIFT算法。为此,提出一种C-GLOH彩色图像配准算法:提取彩色图像各个位置处的颜色不变量,以颜色不变量作为输入图像进行SIFT,其中在构造描述子时,采用改进的GLOH方法,提高了描述子的性能。最后用最近邻法进行匹配。实验结果表明,提出的C-GLOH算法较CSIFT算法有更高的匹配精度、效率和更快的运算速度。Image registration （IR） plays an essential role in all image analysis tasks,which aims to geometrically match overlaying images （two or more） of the same scene taken under different conditions.The properties of feature-based approaches make them widely used among many existing IR approaches.But,many feature-based approaches lose color information by using the gray images as the input images.CSIFT takes the color invariance as the input images,which applies the Color information to image registration successfully.However its speed,efficiency and robustness cannot hold a candle to SIFT.This paper addresses this problem and proposes an C-GLOH approach.The proposed approach builds the SIFT descriptors in a color invariant space,like CSIFT.A improved GLOH method is used to generate the descriptors.The matching result shows that,this approach can stand the comparison with CSIFT.

关 键 词：彩色图像配准 SIFT 颜色不变量 GLOH 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]