针对航空影像密集匹配生成点云数据边界模糊的问题,提出了一种基于DSM灰度影像矢量边界与DEM无约束D-三角网嵌套生成具有精确边界的建筑物表面模型的方法。通过逐点内插法建立实验区点云数据的DSM深度影像图;根据计算机视觉中的边缘检测算子,提取深度影像中建筑物的准确边界;建立DEM的无约束D-三角网,将准确建筑物边界作为硬边界嵌入三角网中,最终将建筑物三角网和地面点三角网拼合,生成“纯净”建筑物表面模型。实验结果表明,优化后的建筑物高度和平面信息无精度损失,该方法有较强实用性。 Aimed at the ambiguity of point cloud data generated by dense matching of aerial images, a method of building a model of building surface with precise boundaries based on DSM grayscale image vector boundaries and DEM unconstrained D-triangulation nets was proposed. Based on the edge detection operator in computer vision, the accurate boundary of building in depth image was extracted. The unconstrained D-triangulation of DEM was established, which would be accurate Building boundaries are embedded as hard boundaries in the Triangulation and eventually the Triangulation of Buildings and Ground Triangulation are combined to create a “Pure” building surface model. Experimental results show that there is no loss of accuracy for the height and plane information of the optimized building, and the method has strong practicability.