摘要
为了解决城市三维重建中单体化困难、数据量庞大以及泊松重建表面起伏、锐利特征丢失的问题,提出一种城市建筑物三维重建方法。使用基于倾斜摄影、Patch-based Multi-view Stereo(PMVS)和泊松表面重建所得密集三角格网作为输入。重建的流程分为两个步骤:建筑物提取和多边形3D模型重建。建筑物提取的方法是:从特定高度切割三角格网得到建筑物的外包围轮廓,进而利用轮廓分割出建筑物。多边形3D模型重建的过程是:从初始高度以指定步长切割三角格网得到轮廓,然后精简轮廓,最后建立起3D模型。实验使用单栋建筑物和大规模城市场景两种数据,实验结果表明算法表现出较好的性能、稳定性和可扩展性,并且具有较高精确性和有效性。
In order to solve the problems in urban 3D reconstruction such as singlarisation difficulty, huge amount of data, and the undulated surfaces and sharp characteristics losing in Poisson reconstruction., we propose a 3D reconstruction method for urban buildings. It uses dense triangular mesh, which is derived based on oblique photography, patch-based multi-view stereo （PMVS） and Poisson surface reconstruction, as the input. The reconstruction process is divided into two steps： the building extraction and the polygon 3D model reconstruction. The building extraction method is as follows： cutting the triangular mesh at a certain height and getting the outer surrounding contour of building, and then segmenting the building by contour. The process of polygon 3D model reconstruction is as follows： cutting the triangular mesh from initial height with a specified step length to get the contours, and then simplifying contours and finally establishing 3D model. The experiments employ two kinds of data： the single buildings and the large-scale urban scene. Experimental results show that the algorithm demonstrates good performance, stability and scalability, and has quite high accuracy and effectiveness.
出处
《计算机应用与软件》
CSCD
2016年第12期188-192,共5页
Computer Applications and Software
关键词
倾斜摄影
建筑物提取
精简
锐利特征
聚类
层次调整
Oblique photography
Building extraction
Simplification
Sharp characteristic
Clustering
Level adjustment
