From Photogrammetry to Virtual Reality: A Framework for Assessing Visual Fidelity in Structural Inspections

Area of Science:

• Civil Engineering
• Computer Science
• Virtual Reality

Background:

• Civil structures face deterioration from natural impacts, necessitating frequent inspections.
• Traditional in situ inspections are costly, risky, time-consuming, and error-prone.
• Virtual Reality (VR) offers a potential alternative for structural inspection platforms.

Purpose of the Study:

• To develop a comprehensive framework for assessing the visual fidelity of VR models for structural inspection.
• To address the critical question of whether VR models accurately reflect real-world structural conditions.

Main Methods:

• Integration of Unmanned Aerial Vehicle (UAV)-based photogrammetry, computer graphics, and web-based VR editing to create interactive VR user interfaces.
• Development of a quantitative visual fidelity assessment methodology.
• Evaluation through image alignment, histogram matching, and pixel-level deviation mapping between VR model renderings and UAV-captured images.

Main Results:

• A novel workflow for establishing interactive VR inspection interfaces was introduced.
• A robust methodology for quantitatively assessing VR model visual fidelity was proposed and validated.
• Case studies on a historic stone arch bridge and a campus steel building demonstrated the framework's effectiveness.

Conclusions:

• The study provides a validated framework for assessing the visual fidelity of VR models in structural inspection.
• This research lays a foundation for advancing VR-based structural inspection technologies.
• The findings support the use of VR as an accurate and reliable tool for evaluating structural integrity.