Extraction of Interior and Exterior IFC Building Surfaces by Visibility-Based Mapping for Point-Cloud-to-BIM Comparison

Abstract

This work presents an element-consistent method for extracting interior and exterior surfaces from Industry Foundation Classes (IFC) building models. The approach addresses the geometric and semantic mismatch between terrestrial laser scans representing the as-is state and IFC models describing the as-designed state. Reliable comparison requires geometrically faithful reference surfaces whose patches remain traceable to their source Building Information Modeling (BIM) elements. The proposed automated pipeline preserves the IFC Level of Development (LOD) without element-wise simplification and maintains each element’s GUID throughout processing. Interior room surfaces and exterior shells are formulated within a unified visibility-based framework in which first-hit triangles are identified directly on IFC-derived meshes by ray tracing. The workflow comprises IFC tessellation, storey-wise room-footprint detection, prism-based interior sampling, first-hit triangle aggregation, element-preserving clipping, and analogous exterior visibility-shell extraction. The Python prototype uses IfcOpenShell and Open3D and processes complex building models without manual intervention. The resulting element-aware meshes provide reference geometry for Scan-to-BIM registration, geometric deviation analysis, and digital twin validation.