基于海量遥感影像数据的影像快速校正试验研究  

Experimental Study on Rapid Image Correction Based on Massive Remote Sensing Image Data

作　　者：周翠敏 黄映聪 何月顺 ZHOU Cuimin;HUANG Yingcong;HE Yueshun(Faculty of Earth Sciences,East China University of Technology,330013,Nanchang,PRC;Key Laboratory of Digital Land and Resources,East China University of Technology,330013,Nanchang,PRC;Network and Information Center,East China University of Technology,330013,Nanchang,PRC;Jiangxi Provincial Engineering Laboratory of Radiology Big Data Technology,East China University of Technology,330013,Nanchang,PRC;State Key Laboratory of Nuclear Resources and Environment,East China University of Technology,330013,Nanchang,PRC;School of Information Engineering,East China University of Technology,330013,Nanchang,PRC)

机构地区：[1]东华理工大学地球科学学院,南昌330013 [2]江西省数字国土重点实验室,南昌330013 [3]东华理工大学网络与信息中心,南昌330013 [4]江西省放射性地学大数据技术工程实验室,南昌330013 [5]核资源与环境国家重点实验室,南昌330013 [6]东华理工大学信息工程学院,南昌330013

出　　处：《江西科学》2022年第4期696-703,747,共9页Jiangxi Science

基　　金：国家自然科学基金项目(41872243,41202160);核资源与环境国家重点实验室开放基金项目(NRE1913);江西省放射性地学大数据技术工程实验室开放基金项目(JELRGBDT202002);江西省数字国土重点实验室开放基金项目(DLLJ201916);东华理工大学博士科研启动基金项目(DHBK2018021)。

摘　　要：地面控制点的选取是遥感影像处理过程中的一项基础工作,当待校正影像数据量较大时手动选点的效率较低。为提高地面控制点提取的效率,以台湾省台中市为研究区,对云端瓦片影像数据进行挖掘,测试SIFT(scale invariant feature transform)算法并用它来开展自动提取地面控制点的试验。首先,分别提取目标影像和参考影像的特征点;然后,通过特征点匹配和错误匹配剔除,计算特征匹配点相对于参考影像特征点的地理坐标;最后,生成目标图像与参考图像特征点的地面控制点文件。实验结果表明:SIFT算法能够自动提取地面控制点,且精度达99%以上,实现了用较低分辨率校正较高分辨率的影像,表明采用SIFT算法挖掘云端遥感影像特征,可以智能、快速地提取待校正遥感影像的地面控制点,对遥感影像智能化处理具有重要意义。The selection of ground control points(GCPs)is a fundamental work in remote sensing image processing.When the amount of image data to be corrected is large,the efficiency of manual selection of GCPs is low.In order to improve the efficiency of GCPs selection,scale invariant Feature Transform(SIFT)algorithm was tested in Taichung City,Taiwan Province,and used to carry out the experiment of automatic ground control points extraction.Firstly,feature points of the target image and reference image are extracted respectively.Then,the geographic coordinates of the feature matching points relative to the feature points of the medium reference image are calculated by feature matching and error matching.Finally,ground control point files of target image and reference image feature points are generated.Experimental results show that the SIFT algorithm can automatically extract ground control point,and the accuracy is above 99%,and achieve the high resolution image with low resolution correction,SIFT algorithm show that the mining characteristics of remote sensing image,the clouds can be smart,to quickly extract for calibration of remote sensing image ground control point,the intelligent remote sensing image processing is of great significance.

关 键 词：数据挖掘 SIFT 特征提取 地面控制点 瓦片影像 

分 类 号：P237[天文地球—摄影测量与遥感]