摘要
实现影像间自动配准的关键在于快速提取足够数量且精度高的匹配点。提取特征点的方法主要有基于特征和基于灰度两种。经过大量的实验证明,sift算法提取特征点,从粗略的提取特征点,再通过四步对已提取的特征点进行提纯,最后利用精度较高的匹配点进行计算影像间的变换参数,实现配准的自动化。
The key for realizing automatic image registration is to quickly extract a sufficient number of matching points with high precision.There are two methods to extract the feature points: based on feature and based on gray scale. Through a large number of experiment, it is proved that the shift algorithm is the best one. After roughly extracting feature points, purify the extracted feature points through four steps, Finally calculate the transformation parameters between images by using the matching points with high precision, so as to achieve automatic registration.
引文
[1]宋佳乾,汪西原.改进SIFT算法和NSCT相结合的遥感图像匹配算法[J].测绘通报,2018,(09):34-38.
[2]罗通.Harris和Sift算法在不同尺度下提取特征点分析[J].信息系统工程,2018,(08):161-162.
[3]张正源.基于SIFT算法的遥感图像配准系统设计与应用[J].同煤科技,2017,(03):15-18.
[4]闻泽联.基于点特征的遥感图像配准方法研究[D].西安电子科技大学,2017.
[5]杨欢,石俊生,字崇德,李希才.一种改进的SIFT图像配准算法[J].云南师范大学学报(自然科学版),2017,37(03):43-49.