Processing Digital Images for Crack Localization in Reinforced Concrete Members

Abstract

Cracks are among the most frequent types of damage occurring in concrete structures. The structural inspection often requires application of non-destructive techniques for localization of damages, and for validation of the structural integrity. Traditionally, cracks are localized and measured using crack width templates or microscopes, and consequently the crack pattern is transferred to a drawing sheet manually. These operations imply a high level of imperfection, subjectivity of judgment, furthermore they are time-consuming. In the engineering practice, digital image analysis systems can be implemented for reliable detection of concrete surface cracking. In this paper, such procedure is proposed. Images obtained by Digital Image Correlation technique are used for the crack localization. The image processing is performed in two steps. First, the image is modified to achieve strictly horizontal position for the purpose of removing effect of perspective and shape deformation. Ambient noise is also reduced. Subsequently, the vertical shape of cracks is used in order to localize their position. The Agglomerative Hierarchical Clustering Technique is used at the second analysis step for identifying the “cracking pixels” (projections) that closely resemble one another. The proposed algorithm can be applied to datasets of the images generated at different loading levels for the purpose of producing a diagram that represents evolution of the crack distances with increasing load. It is illustrated using the experimental data obtained by the authors.