Code for "Semi-Automated Indoor Geometry Reconstruction for Daylight Simulation"

<strong>Note: There is a separate repository that contains the I/O data for evaluation. This is mentioned in the references. </strong><br><strong>Abstract:</strong> This study presents a semi-automated pipeline for reconstructing indoor geometries from point cloud data for daylight simulation. The pipeline generates watertight models of permanent architectural surfaces with window boundaries through three steps: (1) preprocessing, (2) permanent structure reconstruction, and (3) window boundary extraction. The pipeline was evaluated in four rooms of varying complexity against manually reconstructed models, with daylight availability and glare simulations performed in <em>Radiance</em>. Daylight availability results show absolute errors below 10% for UDI, with mean TAI percentage errors within 18% for rooms with rectangular windows and up to 44% for those with non-rectangular windows. The DGP error remains under 4%, and the modelling time does not exceed 5 minutes in any scenario. The approach enables rapid generation of simulation-ready models with acceptable accuracy for CBDM.**注意**：用于评估的输入输出（Input/Output，简称I/O）数据集存放于独立仓库中，相关说明已在参考文献中提及。 **摘要**：本研究提出了一套面向日光模拟的点云数据室内几何结构半自动重建流水线。该流水线通过三个步骤生成带有窗边界的永久建筑表面水密模型：（1）预处理、（2）永久结构重建、（3）窗边界提取。本研究在四间复杂度各异的房间中对该流水线进行验证，以手动重建模型作为参照组，并使用专业光学模拟软件*Radiance*完成日光可用性与眩光模拟。日光可用性结果显示，有效日光照度（Useful Daylight Illuminance，简称UDI）的绝对误差低于10%；对于带有矩形窗户的房间，平均总照度超出阈值时间（Time Above Illuminance，简称TAI）的百分比误差在18%以内，而非矩形窗户的房间该误差最高可达44%。日光眩光概率（Daylight Glare Probability，简称DGP）的误差始终低于4%，且所有场景下的建模时长均不超过5分钟。该方法可快速生成满足模拟需求的模型，在基于气候的日光建模（Climate-Based Daylight Modelling，简称CBDM）中具备可接受的精度。