【摘要】 针对无人机倾斜影像刺点工作的高人工干预和自动化算法鲁棒性不足的问题，提出了一种利用空间相似性拓扑约束的自动化控制点刺点方法。首先，利用传统影像纠正与核线约束完成少量控制点辅助刺点的初步检测；然后，利用空间相似性拓扑约束预测未知控制点空间坐标，自动生成尺度自适应图斑搭配交叉验证机制完成高可靠的区域性目标检测；最后，利用线段检测进行角点提取，在目标检出框内自动提取可靠的亚像素级角点。该方法从摄影测量基本原理出发，利用空间拓扑关系实现高效可靠的自动化检测，利用控制点几何特性完成稳定可靠的角点检测，实现低资源消耗、高效率、高可靠的摄影测量自动化刺点。更多还原

【Abstract】 Addressing the issues of high manual intervention and the lack of robustness in automated algorithms for the photogrammetric tie-pointing work of UAV oblique imagery, this paper proposes an automated control point tie-pointing method based on spatial similarity topological constraints. The method firstly utilizes traditional image correction and kernel line constraint methods to semi-automatically detect a small number of reference control points. Then, based on spatial similarity and topological constraints, the method predicts the spatial coordinates of unknown control points, automatically generates scale-adaptive patches, and utilizes a cross-validation mechanism to achieve reliable regional target detection. Finally, the method uses line detection to extract corner points and automatically extract efficient and reliable sub-pixel-level corner points within the target detection boxes. The entire paper implements a highly automated and efficient control point detection workflow based on traditional methods.更多还原