福州大学学报（自然科学版）2017，Vol.45Issue(1)：121-126,6.DOI:10.7631/issn.1000-2243.2017.01.0121
一种基于仿射变换的SIFT误匹配剔除方法
A method of mismatch elimination based on affine transformation
摘要
Abstract
Aiming at the problem that a great deal of mismatch exists in the process of scale-invariantfeature-transform(SIFT)-based matching,a new method based on affine transformation is proposed.Firstly,the coefficients of the affine transformation are computed by means of the least square method according to initial matched point pairs resulting from rough matching carried out by the SIFT algorithm.Then the matching relationship of point pairs with larger RMSE is eliminated by continuous iteration.The experimental registration results of TM image of Changting County in different time phase show that the proposed method is simple and efficient,and the registration accuracy is better than that of the widely used RANSAC method.关键词
误匹配对剔除/SIFT/仿射变换/遥感图像配准/长汀县Key words
mismatch elimination/scale invariant feature transform/affine transformation/remote sensing image registration/Changting County分类
信息技术与安全科学引用本文复制引用
吴丹力,毛政元..一种基于仿射变换的SIFT误匹配剔除方法[J].福州大学学报（自然科学版）,2017,45(1):121-126,6.基金项目
“十二五”国家科技支撑计划项目(2013BAC08B02-01) （2013BAC08B02-01）
