Automated camera positioning for digital twinning : incorporating 360° imaging, differential rendering, and 3-D spatial data

Abstract

Efficient and accurate camera positioning is crucial in dynamic environments, particularly within the construction and facilities management industry. This study presents an innovative method that combines 360° imaging, differential rendering, and 3-D spatial data to achieve precise camera positioning. Initially, Global Navigation Satellite Systems (GNSS)-based positioning is used, which is then refined using a 3-D spatial model. Synthetic views are generated for comparison with real-world panoramas captured by a 360° camera. By minimizing discrepancies through differential rendering—specifically by comparing material and color attributes of buildings—we enhance camera positioning accuracy and orientation. Experimental results demonstrate the effectiveness of our method across various environments, achieving positioning errors as low as 0.84 m in close proximity to structures and maintaining consistent accuracy across different scenarios. Our approach leverages surrounding buildings and employs level of detail (LOD) weighting, outperforming existing techniques, including smartphone-based positioning, weighted-least-square methods, and the 3D-mapping-aided algorithm. This research addresses key industry challenges, offering substantial time and labor savings, and provides a solid foundation for future advancements.