ABSTRACT
Infrastructure is deteriorating, requiring more efficient inspections. In expressway tunnels, mobile inspection vehicles are promising, as they enable monitoring while moving. Regular monitoring is vital since components like jet fans, lighting, and cables, secured by screws and nuts, are attached to the tunnel lining. These overhead attachments pose a falling risk; however, frequent close-up inspections are difficult due to road closures. This study examines stereo matching feasibility using images from a single camera on a moving inspection vehicle. Stereo matching estimates object distance by aligning feature points in images from two viewpoints. Unlike standard stereo cameras, this system compensates for motion-blur, enabling high-speed image capture at 100 km/h. Experiments used flat and uneven objects on a linear actuator, reconstructing 3D data and evaluating depth accuracy. This method was applied to tunnel data at 100 km/h, analyzing nuts, metal plates, and cables. Future developments include multi-directional motion for infrastructure management.
