L0级机载三线阵影像连接点自动提取  

An automatic tie point matching method for L0 level airborne three-line-array images

作　　者：杜守基 邹峥嵘[1] 张云生[2,1] 张明磊[1] 

机构地区：[1]中南大学地球科学与信息物理学院,长沙410083 [2]水能资源利用关键技术湖南省重点实验室,长沙410014

出　　处：《国土资源遥感》2017年第1期65-70,共6页Remote Sensing for Land & Resources

基　　金：水能资源利用关键技术湖南省重点实验室开放基金项目“基于近景摄影测量的库岸滑坡监测关键技术研究”(编号:PKLHD201310);国家“863”计划项目“典型地质灾害遥感监测与应急关键技术及应用”(编号:2012AA121301);国家自然科学基金项目“自适应三角形约束的多角度影像多基元匹配方法”(编号:41201472)共同资助

摘　　要：针对L0级机载三线阵影像数据量大以及变形大的问题,提出一种基于SIFT和相关系数法匹配相结合的连接点提取方法。首先,建立金字塔影像后,在影像金字塔顶层采用SIFT算法匹配同名点;然后,基于金字塔匹配传播以及相关系数法实现连接点的传递;最后,通过POS数据几何约束剔除可能的误差,并进行分布优化。3个条带ADS40影像的试验结果表明:与常规连接点提取方法相比,该文方法匹配正确率提高6%以上,且连接点分布更均匀,可以满足自动空三的需求。To tackle the problem of the large quantities and serious deformation of L0 level airborne three - line - array images, this paper proposes a tie point matching method based on SIFT algorithm and correlation coefficient. Firstly, pyramid image is generated and SIFT algorithm is used to match initial corresponding points on the top - level of pyramid image. Then tie points are propagated through pyramid images via correlation coefficient matching method. Finally, possible match error is removed based on geometric constraint of POS data, and distribution of tie points are optimized. Three stripes of ADS40 images were used for experiments. Compared with the conventional image tie point transfer method, the proposed method can improve the ratio of correct matches by more than 6% and the tie points are well- distributed.

关 键 词：三线阵影像 影像匹配 SIFT特征 相关系数法 金字塔影像 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]