期刊文献+

基于倾斜影像的三维城市自动建模 被引量:28

Automated three-dimensional urban modeling and accuracy evaluation using terabyte oblique aerial images
原文传递
导出
摘要 针对传统的基于三维GIS和CAD三维建模技术存在人工干预多、建模速度慢、成本高,不适用于大规模城市三维建模等问题,基于激光点云数据建模技术无法直接获取刚性模型纹理,建模成本高等问题,对目前应用较为广泛的基于倾斜影像三维建模技术的技术要点进了研究。该技术以倾斜多视影像为基础,借助于多视影像特征提取与定向技术、密集匹配与三维重建技术、纹理贴图技术进行三维建模。该方法综合运用计算机视觉技术、摄影测量技术,在保证三维模型定位精度高、纹理细节好的同时,节省人力物力,适用于大规模城市三维建模。 Traditional geographic information system and computer aided design based three-dimensional urban modeling approaches need intensive human interaction,human modeling operation and material resources.These approaches are quite expensive and time consuming.Furthermore,the scanned laser point-clouds do not have the rigid texture information.Therefore,these approaches are ill-suited for threedimensional urban modeling in large area of the city.To solve these problems,oblique aerial image acquisition technology is researched.Based on large-scale datasets of oblique aerial images,in this paper,we present an automated novel approach for high-fidelity three-dimensional urban modeling using oblique aerial images in the scale of terabytes.The algorithms of image feature extraction and orientation,dense matching,3D modeling and texture mapping are integrated in this automated approach.Also,we developed several novel ideas integrating state-of-the-art methods in the field of computer vision and photogrammetry.Our approach can not only achieve high-fidelity texture and high-accurate 3Dpositions,but also handles large-scale images in an automated manner with less manpower.To sum up,this approachwill play an important role in three-dimensional urban modeling in the future.
出处 《测绘科学》 CSCD 北大核心 2017年第1期124-130,149,共8页 Science of Surveying and Mapping
基金 千人项目(Y474161) 国家国际科技合作计划项目(2010DFA92720)
关键词 倾斜影像 三维建模 特征提取 密集匹配 纹理贴图 oblique aerial images 3D urban modeling feature extraction dense matching texture
  • 相关文献

参考文献8

二级参考文献86

共引文献228

同被引文献217

引证文献28

二级引证文献168

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部