国土资源遥感Issue(3)：36-41,6.DOI:10.6046/gtzyyg.2015.03.07
基于势函数点分布调整的SIFT图像配准算法
Image registration algorithm based on SIFT and potential function adjusting location of points
摘要
Abstract
Scale invariant feature transform ( SIFT ) is a popular feature extraction algorithm that has applied to remote sensing image automatic registration; nevertheless, there exists a problem in the remote sensing image automatic registration based on SIFT algorithm, i. e. , the distribution of feature points is always nonuniform. An automatic image registration algorithm based on potential function model is presented in this paper, which can solve the problem of optimizing nonuniformity in feature point distribution in SIFT. By adjusting the threshold of SIFT, the number of matching points is promoted. The algorithm can optimize the uniformity in feature point distribution by potential model function in molecular mechanics, and make the low-precision feature point to the sparse area of feature points. Then it revises local mutual information to improve matching point accuracy, so as to realize a high quality ( uniform space distribution, high accuracy of Sub-Pixel registration) automatic image registration finally.关键词
尺度不变特征转换(SIFT)/势函数/特征点分布/局部互信息Key words
scale invariant feature transform ( SIFT)/potential function/feature point distribution/local mutual information分类
信息技术与安全科学引用本文复制引用
孙彬,边辉,王培忠..基于势函数点分布调整的SIFT图像配准算法[J].国土资源遥感,2015,(3):36-41,6.
