Application of Computer Vision Technology to Structural Health Monitoring of Engineering Structures

Automatic Bridge Inspection And Evaluation System

Traditional ways of inspecting bridges mainly depend on manual works or complicated and expensive sensors (Hu et al., 2013). During the inspection, bridge inspection vehicles or vessels are usually applied which is costly and has an adverse impact on traffic. The automatic bridge inspection and evaluation system is a technique combining the unmanned aerial vehicle (UAV) technique with the computer vision technique (Zhang et al., 2012; Ellenberg et al., 2016). This technique aims at solving the problems encountered by the traditional bridge inspection methods including difficulty of operation, existence of blind zones, high cost of inspection, massive time-consumption, and adverse impact on traffic.

During the operation of monitoring, this system obtains images of bridges by visual sensors on-board (Morgenthal et al., 2014), analyzes the images by computer software, and determines accordingly if there are bridge diseases such as cracks, damages, exposure of reinforcing bars, and separation of bearings. Also, it determines the condition of a bridge by the changes of displacement between different parts of the bridge. For example, corrosion of steel bridge can be evaluated by color lump distributions on the bridge.

Brief Introduction of the System

The automatic bridge inspection and evaluation system is consisted of the UAV system and the vision system. The UAV system carries the visual sensors to target positions and makes the cameras to grab the images for analysis. The computer vision system is mainly consisted of high-definition cameras and relevant software. After reaching the target position, the camera starts to grab images of the bridge, and then the system deals with the images using image identification technology for recognizing cracks, damages, exposure of reinforced bars, separation of bearings or other bridge illnesses. To illustrate how the system works, crack identification is demonstrated as below. The main procedure is shown in Figures 1 and 2.

Figure 1.

Main procedure of the system

Figure 2.

Image processing procedure