Spatial augmented reality (SAR)-based 3D BIM visualization through semi-automated geometric calibration of a projector-camera system

Abstract

Spatial augmented reality (SAR) enables shared, projection-based building information modelling (BIM) visualization on physical mock-ups, but its use in architectural workflows is limited by complex projector–camera calibration procedures. This study presents a semi-automated SAR–BIM projection system that reorganizes established computer-vision and projective-geometry techniques into a lightweight computational calibration framework tailored to early-stage design review. The method requires a single reference-surface alignment, followed by automatic surface-specific homography calculation by applying marker detection, avoiding full PROCAM calibration. The system was evaluated using single-volume and multi-volume architectural mock-ups under different projector–camera configurations. Results show projection alignment errors within 3.78–8.53 pixels, corresponding to approximately 1.13–2.56 mm on physical model surfaces, which is sufficient for tasks such as massing evaluation and façade articulation. The calibration process typically completes within 0.3–1.0 s, enabling rapid recalibration during iterative design exploration. A qualitative comparison with conventional PROCAM approaches and commercial projection-mapping tools highlights reduced setup time and lower user effort. The results demonstrate the system’s usability and accuracy, highlighting its potential as a user-friendly solution for collaborative exterior design reviews and identifying avenues for future development.