测绘通报 ›› 2017, Vol. 0 ›› Issue (10): 115-119.doi: 10.13474/j.cnki.11-2246.2017.0327

• 技术交流 • 上一篇    下一篇

一种结合拓扑信息和SIFT特征的多源遥感影像自动匹配方法

杜春鹏1,2, 李景山1   

  1. 1. 中国科学院遥感与数字地球研究所, 北京 100094;
    2. 中国科学院大学, 北京 100000
  • 收稿日期:2017-04-14 修回日期:2017-05-31 出版日期:2017-10-25 发布日期:2017-11-07
  • 通讯作者: 李景山。E-mail:jsli@ceode.ac.cn E-mail:jsli@ceode.ac.cn
  • 作者简介:杜春鹏(1991-),男,硕士,主要从事高性能遥感卫星数据地面预处理研究。E-mail:ducp@radi.ac.cn

A Multi-sensor Remote Sensing Image Automatic Matching Method Based on Topological Information and SIFT Features

DU Chunpeng1,2, LI Jingshan1   

  1. 1. Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100094, China;
    2. Chinese Academy of Sciences, Beijing 100000, China
  • Received:2017-04-14 Revised:2017-05-31 Online:2017-10-25 Published:2017-11-07

摘要: 基于单一特征的匹配办法在多源遥感影像匹配中往往不适用的问题,提出了一种结合拓扑信息和SIFT特征的自动多源遥感影像匹配方法。该方法首先在两幅影像中使用SIFT算法在尺度空间上提取特征向量,其次对这些特征点使用最近邻提取1:N的多个可能的匹配点对,然后结合位置信息和拓扑信息对这些可能的匹配点对进行剔除,并使用RANSAC方法剔除粗差,最终得到同名匹配点。试验结果表明,相比于计算机视觉领域常用的SIFT算法,本文方法可有效地提高匹配正确率,并获得更多正确的同名点。

关键词: 多源影像配准, 拓扑信息, SIFT

Abstract: Based on the question of matching method of single feature matching in multi-source remote sensing images are often not ideal, a matching method for combination of topology information and SIFT automatic feature of multi-source remote sensing image is proposed in this paper. The method first in two images using the SIFT algorithm in scale space to extract feature vectors, then these feature points using the nearest neighbor the extraction of 1:N multiple possible matching points. Secondly, the matching points are eliminated by the combination of location information and topological information, and the double edge matching strategy and the RANSAC method are used to eliminate the coarse tea. The experimental results show that compared with the SIFT algorithm commonly used in the field of computer vision, the proposed method can effectively improve the matching accuracy and obtain more correct points of the same name.

Key words: multi-sensor image registration, topological information, SIFT

中图分类号: