Automated crack pattern measurements for rapid post-earthquake safety assessment
Current manual post-earthquake assessment practices are time-consuming, primarily qualitative and costly, thus preventing re-occupancy and first response operations from happening in a timely manner. Automating the current manual practices can reduce the overall assessment time. In that regard, methods have been created for detecting visible damage on structural elements from visual data. However, little work has been found in automatically retrieving useful damage properties needed to estimate the structural elements damage states. This paper presents the authors' recent work in automatically retrieving crack properties, comprehensively represented in the form of the surface crack pattern, and further the classification of the surface crack pattern according to the standard crack types. Real concrete column surface images were employed to validate the work presented in this paper. For an image, the automatically retrieved crack pattern properties are compared with the crack pattern properties which were retrieved manually. From these results, we can see that the method in crack pattern classification presented in this paper is highly effective.