Minimal Paths for 3D Crack Detection in Concrete

Concrete is a commonly used construction material for buildings, bridges and roads. As safety is very important in such constructions, the investigation of damage processes in concrete is a highly relevant topic. For instance, the early detection of cracks can prevent the collapse of a bridge. Thus, the necessity of automated crack detection and segmentation arises. We generalize and modify a 2D method for crack detection in road pavement images to 3D. It is based on modeling the image as a graph and searching for minimal paths therein. The proposed 3D method is evaluated on synthetic crack images and applied to a 3D computed tomography image of real concrete.