ABSTRACT
Using real-time crack width calculation with Canny edge detection, errors may occur in the crack itself due to factors such as the direction, continuity, and shape of the crack. On the other hand, according to the tests, human factors such as the observer’s distance from the crack, the observing angle, and the observer’s movement speed also influence the measured crack width. Therefore, this paper will conduct further experiments on human factors through collecting crack width data using Augmented Reality (AR) headsets. If these human factors can be systematically addressed, more accurate data can be obtained, making it easier for observers to use the system and reducing the impact of terrain constraints. Focusing on the error caused by distance and angle, there is a positive correlation between the error and both factors. Additionally, when further investigating the distance, there is an exponential relationship between the pixel unit of the crack and the distance. Various factors contribute to errors in outdoor crack measurements, such as lighting conditions, observation angles, and distance limitations. To initially exclude these factors, this paper focuses on indoor experiments, with the expectation that the findings could be applied to future outdoor observations.
