期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2010
卷号：XXXVIII - Part 1
出版社：Copernicus Publications
摘要：Nowadays, with the rapid development of 3D building model based applications, there is an urgent demand to develop automatic techniques for integrating 3D outdoor building models with 2D and 3D indoor information to produce semantically, geometrically and topologically correct 3D building models. 3D terrestrial laser scanning (TLS) data provides the accurate 3D geometric information, whereas 2D floor plan has useful semantic facade and indoor information about a building. Therefore, the two datasets are complementary and the integration of these two datasets not only could provide a way to integrate 2D and 3D CAD and GIS data, but also can resolve many practical problems in 3D building modeling. As a first step, this paper presents a semantic and geometric information integrated point matching based method for automatic co-registration of 3D TLS points and 2D floor plans. In order to find the correspondences between the two datasets, the 3D-to-2D registration problem is converted to point matching by coding the invariant geometric and semantic context information into a sequence of points using a defined shape description. Then a similarity score formula is proposed to find the initial matching points and after iterative refinement, all the potential corresponding points are found and used to calculate the transformation. The method was tested using real datasets and produced successful results with high accuracy, which demonstrates the feasibility to register 2D floor plan with 3D TLS data
关键词：Registration; TLS; Point Cloud; CAD; Floor Plan; Automation; Point Matching
