基于影像匹配的点云模型重建技术对比  

Comparison of point cloud reconstruction techniques based on image matching

作　　者：汤念 余建军 徐红波 郑亮[3] 彭瑞 Tang Nian;Yu Jianjun;Xu Hongbo;Zheng Liang;Peng Rui(Beijing University of Civil Engineering and Architecture,Beijing 102616,China;Zhejiang Academy of Surveying and Mapping,Hangzhou 310000,China;CCCC Second Highway Consultants Co.,Ltd.,Wuhan 430056,China;Zhejiang Research Institute of Territory Spatial Planning,Hangzhou y 310000,China)

机构地区：[1]北京建筑大学,北京102616 [2]浙江省测绘科学技术研究院,杭州310000 [3]中交第二公路勘察设计研究院有限公司,武汉430056 [4]浙江省国土空间规划研究院,杭州310000

出　　处：《工程勘察》2021年第6期62-67,共6页Geotechnical Investigation & Surveying

基　　金：北京建筑大学研究生创新项目(PG2020076)。

摘　　要：当前基于倾斜影像密集匹配生成的点云包含丰富的颜色及纹理信息,但是缺乏针对影像密集匹配点云密度、精度等信息进行分析的相关文献,本文通过ContextCapture、COLMAP、COLMAP+OpenMVG、OpenMvg+OpenMVS四种不同方式对倾斜摄影影像进行匹配得到密集点云,分析了不同方式的适用条件以及成果影像匹配点云的差异,最后比选出最佳技术方法,期望为同行提供参考。The current point cloud generated by techniques based on the dense matching of oblique images contains rich color and texture information,but the relevant literature is lack of the analysis on the density and accuracy of the dense image matching point cloud. This article presents how to match oblique photographic images to obtain a dense point cloud by four different methods: ContextCapture,COLMAP,COLMAP +OpenMVG and OpenMvg+OpenMVS. The applicable conditions of the different methods and the differences in the matching point clouds of the result images are analyzed at the same time. Finally,the best technical method is selected and expected to provide a reference for the peers.

关 键 词：点云 倾斜摄影 影像匹配点云 点云精度评估 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]