Author(s): Ki-In Bang
Linked Author(s):
Keywords: Drone BIM Registration Facility Management
Abstract: Introduction River facilities are exposed to dynamic environmental conditions, making it critical to ensure their safety and operational integrity. Image-based facility management using drones encounters challenges, especially in GNSS (Global Navigation Satellite System) -restricted environments. This study proposes an innovative approach to achieving precise image registration between drone-acquired imagery and 3D models. The method aims to provide accurate and efficient management solutions for river facilities. Methodology The proposed system integrates advanced multi-sensor data acquisition and processing technologies and comprises the following four key steps: 1. 3D Model Generation A detailed 3D model of the target river facility is constructed within a virtual environment using structural data. This model serves as a reference point for remote diagnostics and safety evaluations. 2. Virtual Image Generation Multi-sensor systems, including GNSS/IMU and imaging devices, are used to capture location and orientation data. Based on this data, virtual images are generated to replicate real-world perspectives, enabling comparative analysis with actual drone-acquired imagery. 3. Error Analysis Discrepancies between the virtual and real-world images are identified by extracting and comparing key features, such as outlines and contours. Projection transformation methods are applied to calculate alignment errors with high precision. 4. Alignment Correction Identified errors are iteratively corrected to improve alignment between real-world images and the virtual model. Transformation matrices are utilized to adjust the coordinates of captured images to the virtual environment, minimizing alignment errors and ensuring a reliable foundation for safety assessments. Implementation and Application This methodology was tested on various river facilities, demonstrating its effectiveness in environments with limited GNSS signals or challenging accessibility. By integrating multi-sensor data, the system ensures accurate remote safety assessments while reducing manual interventions. Applications include monitoring bridge substructures, pipelines, and other critical installations exposed to complex environments. Conclusion The proposed multi-sensor-based approach offers a robust solution for the safety assessment of river facilities. By integrating virtual models with real-world data, this method enhances precision, reduces GNSS dependency, and supports real-time diagnostics. The approach aligns with the growing demand for scalable and automated solutions in remote monitoring and infrastructure management.
Year: 2025