Accuracy assessment of detecting cracks on concrete wall at different distances using unmanned autonomous vehicle (UAV) images

Abstract

Distance plays an essential role in detcrmining crack detection. Without proper or fixed value of length, the time taken to capture the image of a crack may increase. It is essential to know the capability of Unmanned Autonomous Vehicle (UAV) to capture images for detecting cracks at different distances as it may be useful for building inspection, especially as it shortens the time taken to inspect cracks on buildings. This paper aims to analyze the accuracy of the UAV image in detecting cracks at different distance measurements qualitatively and quantitatively. The observed UAV images of the concrete wall starting from 1m until 6m distances were obtained. The images were processed using Agisoft software, while the edge detection algorithm of the Digimizer software was used for the crack width detection process. It was found that in detecting cracks at different distances, the hairline crack was visible at 3 meters, the narrow crack at 5 meters, and the medium crack at 6 meters. The relationship between distance and size of crack allows for a better understanding on how distance affects the detection of the crack which may be useful for building inspections or in the engineering sector.