【作者】 华开成； 龙毅；

【机构】 广东惠清高速公路有限公司； 南京师范大学；

【摘要】 计算机视觉中的泊松(Possion)构网是实现三维物体表面重建的优秀算法。本文将其引入到航空摄影测量中,利用泊松构网构建数字表面模型(Digital Surface Model,DSM),从而实现新模型下正射影像(DOM,Digital Orthophoto Map)的生成。利用两组包含有高架桥场景的无人机影像进行实验,通过与常规的Delaunay三角网化DSM构建下DOM生成做对比,发现泊松构网生成的DSM更平滑,在高差突兀区域(特别是高速公路的高架桥边缘)生成的DOM效果更好,通过对墩柱中系梁边缘进行矢量化提取,拟合出墩柱中系梁轴线,对高墩柱轴线偏位进行了检测,取得了较好的效果。更多还原

【Abstract】 The Possion network reconstruction in computer vision is an excellent algorithm for surface reconstruction of three-dimensional objects. In this paper, it is introduced into aerial photogrammetry, and the Digital Surface Model(DSM) is constructed by Poisson network to realize the generation of the Digital Orthophoto Map(DOM) under the new model. Doing experiment with two sets of UAV images containing viaduct scenes,by comparing the DOM from the DSM generated by conventional Delaunay triangulation, it is found that the DSM generated by the Poisson network is smoother, and the DOM at the area(especially the edge of the viaduct on the highway) with abrupt height difference is better, Through the vectorization extraction of the edge of the beam in the pier column, the axis of the beam in the pier column is fitted, and the deviation of the axis of the high pier column is detected, and good results are obtained.更多还原