摘要: |
This project will develop a novel fatigue crack inspection technique by integrating computer vision-based motion tracking and augmented reality (AR) to help bridge inspectors accurately detect, track, and document fatigue cracks in the filed in a cost-effective and safe manner. Work in Stage 1 will focus on developing core computer vision and AR algorithms for fatigue crack inspection. Camera compensation strategies will be devised to remove extraneous motion in the video introduced by the camera through geometric transformation and image registration. The camera motion compensation algorithm will be integrated with surface motion tracking and analysis algorithms to enable detecting and quantifying fatigue cracks in the video. Computation time of video processing will be accessed and the developed algorithms will be optimized to minimize processing time through strategies for video stabilization and feature tracking. Communication and interfaces for inspector-crack holographic experiences through computer vision will be established. Application will be developed to generate hologram with crack identification results. Anchoring and surface deployment strategies will be developed for holographic images, and the overlay of the hologram will be tested on the structure surface being inspected. Work in Stage 2 will focus on laboratory and field validations for the developed fatigue crack inspection tool. A parametric study will be performed to quantify the capability of the crack detection algorithm. A large-scale bridge-girder-to-cross-frame connection will be used to create realistic out-of-plane fatigue cracks for assessing the developed tool, including the crack detection algorithm and the interactive AR inspection applications. A full-scale highway steel bridge will be used for field validation. In addition to validating crack detection algorithms and AR applications, thresholds of traffic load level for effective crack identification will be established and the impact of lane closure assessed. |