Contactless Displacement Measurement for Vision-Based Structural Health Monitoring Using Optical Flow

Inspection and assessment are necessary steps to maintain and ensure the safety and capacity of critical infrastructure systems such as bridges and other civil structures. Traditional sensors, including accelerometers, strain gauges, and string potentiometers, can be difficult to place in remote, offshore, and complex environments where the structure is very tall or hard to reach. Vision-based systems offer an innovative approach for structural health monitoring because they do not require contact with the target structure. This work proposes a contactless vision-based system for structural displacement measurement with real-time processing potential. Image acquisition and processing was implemented on an NVIDIA Jetson TX2 connected to a CMOS camera using the Lucas-Kanade-based optical flow method for feature tracking. A series of lab tests studied the performance of the system under a variety of targets, lighting conditions, and motion frequencies to validate the tracking algorithm against a cross-correlation method and a traditional displacement measurement device. Field testing measured the end displacement of a traffic signal mast arm and results show that the system captured movement on a subpixel level. With confirmation of feasibility of the proposed system to provide contactless vision-based displacement measurement through feature tracking, recommended steps for deployment of the system in the built environment for short-term monitoring of bridges or traffic signal structures will be offered.