IEEE Comput Graph Appl. 2021 May-Jun;41(3):48-58. doi: 10.1109/MCG.2021.3069856. Epub 2021 May 7.
The growing demand for building information modeling (BIM) data and ubiquitous applications make it increasingly necessary to establish a reliable way to share the models on lightweight devices. Building scenes have strong occlusion features and the building exterior plays an important role in digital devices with limited computational resources. This allows the possibility to reduce the resource consumption while roaming in outdoor scenes by culling away the interior building data. This article addresses the task of automatic annotation of BIM building exterior via voxel index analysis. We showcase the research of using industry foundation classes (IFC) and other mainstream formats as our input data and proposed an automatic algorithm for annotating the building exterior. Afterward, a practical and accurate voxel index analysis procedure is designed for frequently flawed models. The annotation can be added directly into the original data file under the same IFC standard, avoiding the complex procedure and information loss in semantics mapping between different standards. The final examinations show the robustness of our algorithm and the capability of handling large BIM building models.
对建筑信息模型 (BIM) 数据的需求不断增长，无处不在的应用程序使得在轻量级设备上共享模型成为越来越必要的手段。建筑场景具有很强的遮挡特性，并且在计算资源有限的数字设备中，建筑物的外部起着重要作用。这使得在户外场景中漫游时通过剔除内部建筑物数据来减少资源消耗成为可能。本文通过体素索引分析来解决 BIM 建筑物外部的自动标注任务。我们展示了使用行业基础类 (IFC) 和其他主流格式作为输入数据的研究，并提出了一种用于标注建筑物外部的自动算法。之后，设计了一种针对常见缺陷模型的实用且准确的体素索引分析流程。标注可以直接添加到相同 IFC 标准下的原始数据文件中，避免了不同标准之间语义映射的复杂过程和信息丢失。最终的检验表明了我们算法的健壮性和处理大型 BIM 建筑模型的能力。
