点云信息辅助的航空正射影像自动镶嵌方法

Automated Seamline Detection for Aerial Orthophoto Mosaicking Assisted by Point Cloud Information

航空正射影像镶嵌线的自动化生成能大大减少摄影测量产品生产的人工工作量。目前,仅依赖航空影像本身的自动镶嵌方法难以做到建筑物等障碍地物的自动避让,因此引入DSM、点云、GIS轮廓数据等作为辅助,是解决自动镶嵌问题的有效途径。本文提出了一种新的点云高程与分类信息辅助的双像镶嵌线自动生成方法。先利用点云信息通过动态规划获得绕开障碍地物的初始镶嵌线,然后通过重叠影像的相似度等特性构建代价矩阵,在一定缓冲区内对初始镶嵌线进行优化,从而得到最终镶嵌线。通过宁波测区和斯图加特测区的两组正射影像与点云数据进行测试,证明了本方法可以实现高质量的镶嵌线自动生成,有效绕开建筑物等障碍地物。

The automated generation of seamlines can greatly reduce the manual works for aerial orthophoto mosaicking.However,the avoidance of obstacles like houses and bridges is difficult when generating seamlines without the assistance of additional information such as digital surface model,point-cloud data,or GIS data.In this paper,a two-step method of automated two-view orthophoto mosaicking assisted by the elevation and the classification information of point-cloud data is introduced.First,an initial seamline is generated by a dynamic programming process with the point-cloud.Second,the initial seamline is adjusted within a buffer zone according to the overlapping image information to get the final seamline.The method is tested with two aerial image-pairs collected over Ningbo,China and over Vaihingen,Germany,which demonstrates the high quality of the seamlines automatically obtained by the proposed method.