五方向倾斜航空影像建筑物墙面自动提取方法

Automatic wall extraction from five direction oblique aerial image

针对城市三维建模的立面几何缺失以及倾斜航空影像利用率不高的问题,提出了一种基于倾斜航空影像自动获取精细建筑物立面的方法。该方法先对五方向倾斜航空影像匹配的点云数据在"像方"构建Delaunay三角网,实现目标城区在各方向上的再现;其次,对多视网格进行交叠检测剔除重叠冗余数据;最后,对倾斜非冗余点进行水平投影,采用点密度阈值分割获取建筑物立面精细网格数据。试验结果表明,该方法可实现五方向倾斜航空影像建筑物立面的自动提取。

A method for obtaining the fine facade from the oblique aerial images with abundant redundant information is pres- ented to focus on the lack of geometric facade during 3D city modeling and the low aerial image utilization rate. According to the five direction oblique aerial image matched point cloud, a Delaunay grid is constructed to achieve the target city in all di rections. Then the redundant data is eliminated by overlapping detection in multi-view grid. Finally the real facade grid data is obtained by project the oblique non-redundant point on horizon plane and density threshold segmentation. The experimen- tal results show that this method can effectively realize the facade automatic extraction from five direction oblique aerial im- age building.