沙漠地区无人机影像连接点提取.pdf
2017年测绘科学技术学报2017
第34卷第4期JournalofGeomaticsScienceandTechnology7Vol.34No.4
文章编号:1673-6338(2017)04-0405-06
沙漠地区无人机影像连接点提取
薛武\张永生\于英\赵玲2,李磊1
(1.信息工程大学,河南郑州450001;2.91039部队,北京102400)
摘要:提出了一种纹理自适应的影像连接点提取方法,有针对性地解决沙漠地区无人机影像对比度低、相
似性强、纹理信息不均衡,经常会造成连接点过少甚至提取、匹配失败的突出问题。该方法利用影像金字
塔进行粗匹配,建立影像之间近似单应变换关系;通过对原始影像分块计算灰度共生矩阵的熵来评估影像
纹理指标,并据此实现特征点的自适应提取与匹配。实验表明,该方法能够得到数量较多、分布均匀、可靠
性强的同名点。
关键词:沙漠;连接点;灰度共生矩阵;熵;自适应;并查集
中图分类号:P231文献标识码:AD0I编码:10.3969/j.issn.1673-6338.2017.04.014
TiePointsExtractionofUAVImagesinDesertArea
XUEWu1,ZHANGYongsheng1,YUYing1,ZHAOLing2,LILei1
({.InformationEngineeringUniversity,Zhengzhou450001,China;
2.91039Troops,Beijing102400,China)
Abstract:Imagescapturedindesertareaareoftenwithlowcontrast,repetitivepatternsanduneventexture,which
willresultinscarcetiepointsoranumberofmismatches.Atextureadaptivemethodfortiepointextractionispro
posedfordesertarea.Firstly,theimagepyramidisbuilttoperforminitialmatchingandestablishtheapproximate
homographytransformationbetweentheimages.Then,thetexturelevelisevaluatedbasedonthegreylevelco-
cocurrencematrixoftheimageblocks,andtheadaptiveextractionandmatchingofthefeaturepointsareundertak
en.Experimentsshowthemethodcanobtainconsiderable,relativelyrobustandhomogenouscorrespondingpoints.
Keywords:desert