Toward Building and Civil Infrastructure Reconstruction From Point Clouds: A Review on Data and Key Techniques

Nowadays, point clouds acquired through laser scanning and stereo matching have deemed to be one of the best sources for mapping urban scenes. Spatial coordinates of 3-D points directly reflect the geometry of object surfaces, which significantly streamlining the 3-D reconstruction and modeling of objects. The construction industry has utilized point clouds in various tasks, including but not limited to, building reconstruction, field inspection, and construction progress tracking. However, it is mandatory to generate a high-level (i.e., geometrically accurate, semantically rich, and simply described) representation of 3-D objects from those 3-D measurements (i.e., points), so that the acquired information can be fully utilized. The reconstruction of 3-D objects in a scene of man-made infrastructure and buildings is one of the core tasks using point clouds, which involves both the 3-D data acquisition and processing. There are few systematic reviews summarizing the ways of acquiring 3-D points and the techniques for reconstructing 3-D objects from point clouds for application scenarios in a built environment or construction site. This article therefore intends to provide a thorough review of the state-of-the-art acquisition and processing techniques for building reconstruction using point clouds. It places particular focus on data acquisition and on the strengths and weaknesses of key processing techniques. This review work will discuss the limitations of current data acquisition and processing techniques, as well as the current research gap, ultimately providing recommendations on future research directions in order to fulfill the pressing needs of the intended construction applications in the foreseeable future.