Abstract Scan-to-BIM of existing buildings is in high demand by the construction industry. However, these models are costly and time-consuming to create. The automation of this process is still subject of ongoing research. Current obstacles include the interpretation and reconstruction of raw point cloud data, which is complicated by the complexity of built structures, the vast amount of data to be processed and the variety of objects in the built environment. This research aims to overcome the current obstacles and reconstruct the structure of buildings in an unsupervised manner. More specifically, a novel method is presented to automatically reconstruct BIM wall objects and their topology. Key contributions of the method are the ability to reconstruct different wall axis and connection types and the simultaneous processing of entire multi-story structures. The method is validated with the Stanford 2D–3D-Semantics Dataset (2D–3D-S).

中文翻译：

---

基于点云数据的建筑信息建模墙对象的无监督重建

摘要 建筑行业对现有建筑的扫描到 BIM 的需求很高。然而，这些模型的创建成本高昂且耗时。这个过程的自动化仍然是正在进行的研究的主题。当前的障碍包括原始点云数据的解释和重建，由于建筑结构的复杂性、要处理的大量数据以及建筑环境中的对象种类繁多，因此变得复杂。这项研究旨在克服当前的障碍并以无人监督的方式重建建筑物的结构。更具体地说，提出了一种自动重建 BIM 墙对象及其拓扑的新方法。该方法的主要贡献是能够重建不同的墙轴和连接类型以及同时处理整个多层结构。该方法已使用斯坦福 2D-3D-语义数据集 (2D-3D-S) 进行验证。