Three-Dimensional Mapping of Heritage Structure Using Low-Cost UAV and Terrestrial Photogrammetry Techniques: Case Study of a Heritage Church in Solapur, Maharashtra, India

A common challenge in documenting Indian heritage structures is the lack of comprehensive records, particularly those related to construction drawings. Existing documentation methods typically emphasize exterior surfaces, often neglecting the interiors. These methods also face significant challenges, such as dependence on specialized equipment, the need for skilled personnel, and the difficulty of integrating various data types, underscoring the necessity for more efficient documentation techniques. This study proposes a low-cost procedure integrating unmanned aerial vehicle (UAV) photogrammetry with terrestrial close-range photogrammetry to achieve accurate three-dimensional (3D) documentation. The approach is applied to a 19th-century heritage church in Solapur, Maharashtra. The methodology employs open-source tools for image processing, point cloud generation, and registration using structure from motion and iterative closest point algorithms. The final registered model achieves a root mean square error of 0.0092345 m at 80% overlap and an average length deviation of 2 cm when validated against the terrestrial light detection and ranging-based model and field measurements. These findings demonstrate that the proposed approach, which outlines key surveying phases and a point cloud registration method for nongeotagged images of both interior and exterior surfaces, is effective for heritage structure documentation. To further enhance this documentation process, point cloud model visualization is enabled through WebGL and Three.js using a lightweight Graphics Library Transmission Format, allowing for real-time interaction, measurement, and exploration via web browsers. The proposed procedure demonstrates that scalable, precise, and cost-efficient 3D documentation is achievable, making it well-suited for heritage sites lacking original construction data.