北京测绘2025，Vol.39Issue(5)：631-635,5.DOI:10.19580/j.cnki.1007-3000.2025.05.009
一种面向实景三维建筑物的分层单体化技术
A layered monomerization technology for real-world three-dimensional buildings
摘要
Abstract
The production of monomerized data for real-world three-dimensional(3D)models requires significant human intervention,resulting in high time and economic costs,which makes large-scale application difficult.To address this issue,this paper proposed a layered monomerization technology for real-world 3D models,centered on deep learning-based window object detection.By establishing an artificial intelligence-based window object detection system for texture mapping,the system projected window contours from the texture mapping into 3D space and calculated their central coordinates.These central coordinates were then clustered based on height,and the intermediate position of the two height clusters was used as the layering boundary for the building.The technology was tested on an office building in an industrial park,and the results show that this technology can effectively detect window objects and perform monomerization and layering of the model.It also offers advantages such as fast processing speed,automation,and intelligence,making it widely applicable to real-world 3D monomerization data production,thus improving work efficiency.关键词
实景三维/神经网络/聚类/坐标转换/纹理贴图Key words
real-world three-dimensional(3D)/neural networks/clustering/coordinate transformation/texture mapping分类
测绘与仪器引用本文复制引用
安冠星,寇媛,张正强,申永伟,刘新定,罗建新..一种面向实景三维建筑物的分层单体化技术[J].北京测绘,2025,39(5):631-635,5.基金项目
湖南省自然资源科技计划(湘自资科20230153CH) （湘自资科20230153CH）
