Defect Change Detection in Masonry Structures Using Image-Based 3D Photogrammetry

Abstract

Masonry structures are deteriorating due to environmental and man-made factors, which cause damages such as cracks in the structures. Periodic inspection is required to determine if cracks have changed in appearance over time. In this paper, an image-based system is proposed to detect changes in cracks on the surface of a masonry structure. The system can detect changes in images taken from different viewpoints. In the proposed method, an image-based 3D modeling technique is applied to rectify images taken from different viewpoints via an image synthesis technique. The system deploys a drone to collect images to create a realistic 3D surface model and camera calibration to obtain a reference model. Queried images around the damaged regions are taken at a later time, in which a 3D model from this region is reconstructed and registered onto the reference 3D model. The queried and reference images are synthesized via a 3D model and compared by a threshold-based algorithm to detect changes between the images. Different methods of image synthesis were compared to find the most accurate change masks. It was shown that the proposed system can rectify and detect changes in images accurately, even though images are from different viewpoints and a complex surface.