Surface Cracks Length Calculation in Segmented Images using Image Processing Techniques

Civil infrastructure systems make up a large share of global assets. Hence, frequent crack inspection is crucial in maintaining infrastructures. The conventional methods of examining surface cracks are time-consuming and subjective. However, the vision-based component of unmanned aerial vehicles draws an advantage in replacing the traditional laborious approach. The alternative approach for crack inspection is automation by image processing techniques. This paper introduces a method to calculate the length of surface cracks. The segmented images are the outcomes of semantic segmentation prediction models. The authors developed a software application that allows end-users to provide endpoints of the crack, and the application will calculate this crack segment instantaneously. The introduced algorithm yielded an average error of 4.26 mm and 3.17 mm from the two gathered datasets. The experimental results proved that the developed image processing algorithm for surface crack length measurement could calculate actual surface crack with tolerable errors.