多基准点和基准方向的面阵与线阵影像匹配方法  

A method of plane array image and line array image matching based on multi reference points and reference directions

作　　者：柯涛[1] 陈彪[1] 李从敏[2] KE Tao CHEN Biao LI Congmin(School of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430079, China State Key Laboratory for Information Engineering in Sur- veying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China)

机构地区：[1]武汉大学遥感信息工程学院,武汉430079 [2]武汉大学测绘遥感信息工程国家重点实验室,武汉430079

出　　处：《测绘科学》2017年第8期73-78,141,共7页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(41201482)

摘　　要：针对面阵影像与线阵影像之间存在的尺度、旋转等几何差异所导致的难以高效获得稳定可靠匹配点问题,该文提出了一种稳健的面阵影像与线阵影像匹配方法。算法分为3个阶段:(1)对影像进行尺度、旋转预处理,消除影像间的尺度和旋转差异;(2)用Harris算子提取特征点,进行逐层金字塔匹配,在最底层影像上进行最小二乘匹配获取精匹配点,利用随机抽样一致算法获取准确性较高的匹配点;(3)对准确性较高的匹配点分区,构建4块匹配区域,根据每块区域的匹配点计算重心,以此4块区域的重心作为匹配基准点和基准方向,重约束初始匹配结果。通过多组面阵影像与线阵影像数据实验结果,验证了该方法的可行性和有效性,可为后续空中三角测量及相关应用提供可靠的匹配点。Aiming at the problem that it is difficult to efficiently obtain reliable corresponding points between plane array image and line array image due to scale,rotation and geometric differences,a robust matching method was proposed in this paper,which was divided into three stages：（1）preprocessing images to eliminate scale and rotation difference;（2）extracting feature points using Harris operator and conducting pyramid match layer by layer,and then least squares match was applied to obtain exact match points from the original image.On this set of features,random sample consensus（RANSAC）was run to eliminate wrong matches;（3）dividing remaining higher accurate match points after RANSAC into four blocks and calculating the mean of x,y coordinates according to remaining matching points of each block,striking the center of four blocks,which were used as reference points and reference directions,rebooting the initial matching results.Experimental results showed that the proposed method was feasible and effective,and it could provide reliable homonymy points for the subsequent aerial triangulation and related applications.

关 键 词：面阵影像 线阵影像 基准点 基准方向 Harris特征点 相关系数 分块匹配 

分 类 号：P23[天文地球—摄影测量与遥感]