Vision-Based Damage Detection Using Inclination Angles and Curvature

This paper presents damage detection techniques for structural health monitoring of horizontal structures using computer vision. A technique based on the derivation of curvature from the second order polynomial equations of the deflection curve is introduced. The technique, as well as inclination angles, and the primary deflection data are applied for damage detection on a simply supported laboratory beam subjected to a point load at its midspan. The beam is loaded and unloaded at intact and damaged states. Measurements are obtained with a smartphone. The measurement resolution is 1 mm/px—a relatively low value. Measurements are pre-processed for measurement noise. Results show that damage can be detected using all three responses analysis techniques. The curvature and inclination angle techniques outperform the deflection technique, especially for damage identification