基于SIFT特征的遥感影像自动配准  

SIFT Based Automatic Registration of Remotely-sensed Imagery

作　　者：李晓明[1] 郑链[1] 胡占义[2] 

机构地区：[1]北京理工大学机电工程学院 [2]中国科学院自动化研究所模式识别国家重点实验室,北京100080

出　　处：《遥感学报》2006年第6期885-892,共8页NATIONAL REMOTE SENSING BULLETIN

基　　金：国家自然科学基金资助课题(编号:60375006);山西省自然科学基金(编号:20051032)资助

摘　　要：遥感影像的自动配准是长期以来一直未能很好解决的一个重要问题。本文将视频图像匹配中获得巨大成功的SIFT(Scale Invariant Feature Transform)特征应用于遥感影像的自动配准问题中,并且针对遥感影像的成像特点,给出了一种具体的特征匹配方法。对航空和航天遥感影像在不同的变形、不同的光照变化和不同的分辨率下进行的大量实验表明,该方法具有稳定、可靠、快速等特点。Automatic registration of remotely-sensed imagery is a classical problem, and has not been well solved till now. In this paper, SIFT（Scale Invariant Feature Transform） feature, which has shown great success in computer vision, is introduced into image registration in remote sensing. In addition, we also proposed a feature matching approach based on the specific characteristic of the remote sensing imagery. Numerous experiments have been conducted for both aerial and satellite imageries under various conditions such as geometric distortion, illumination variation and different resolution. The results showed that our matching approach performs well, and is stable, reliable and efficient.

关 键 词：遥感影像 自动配准 SIFT 图像配准 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]