一种可靠的多光谱CCD遥感图像配准方法  

A Reliable Registration Method for Multispectral CCD Remote Sensing Images

作　　者：陈慧津[1] 赵龙[1] 陈哲[1] 

机构地区：[1]北京航空航天大学自动化科学与电气工程学院,北京100083

出　　处：《微计算机信息》2010年第8期1-3,共3页Control & Automation

基　　金：航空基础科学基金(省部级基金);基金申请人:赵龙;项目名称:惯性/地形辅助导航技术研究;基金颁发部门:中国航空工业第一集团公司(20070851010)

摘　　要：多源遥感图像数据的融合和后继应用必须先实现严格的配准。针对CBERS-02B遥感卫星多光谱CCD相机的成像尺寸大、像质差异性小及不同波段图像间的误差范围较固定等特点,提出一种基于小窗口多次匹配的图像配准方法。即在整幅大图像上均匀获取多个小窗口图像与基准图像进行灰度NCC匹配,将多个匹配结果通过剔除误匹配点后求平均的方法来获得具有子像素级精度的配准结果,从而保证配准的可靠性和精度。将该方法用于对多光谱CCD图像的配准实验中,结果表明本文提出的配准方法具有可行性和实用性。In remote sensing applications, accurate registration is very important for image data fusion and other applications.Images acquired by the multispectral CCD camera in CBERS-02B remote sensing satellite have 3 characteristics：the size of image is very large;the discrepancy of images nature is small;their postural distance is in a fixed range.Aimed at these characteristics, a reliable registration method based on small-window and many times of matching calculations is presented.Some small window images symmetrical gotten from a large image do NCC matching calculate with the reference map.Then there are many matching results.Those mismatching points are eliminated by some algorithm.The average of remaining matching results can produce sub-pixel accuracy registration between multispectral images.Real multispectral CCD images of CBERS-02B remote sensing satellite were registered by this registration method in simulation experiments, and the results indicated that this algorithm is feasible and practical.

关 键 词：多光谱 图像配准 遥感图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]