An Accurate Vision-Based Dynamic Propagation Measurement Method for Simple Concrete Cracks

Abstract

To measure dynamic crack evolution under sustained service loads during infrastructure defect inspection and structural condition assessment, a machine-vision-based methodology for high-accuracy measurement of simple crack propagation is proposed in this paper. A dynamic crack propagation characterization framework is first established to number the cracks and trace the cracks across distinct image frames, and an accurate vision-based dynamic propagation measurement method for simple concrete cracks is further developed by integrating the proposed characterization approach. Subsequently, an indoor compression experiment of a concrete specimen was carried out to measure the cracking process with the proposed method, and the dynamic variations in the subpixel-level width and length were calculated for each crack. The cracking initiation, coalescence, acceleration and ultimate failure stage were also identified and recorded. Finally, the accuracy of crack width measurement was validated via the digital image correlation (DIC) method. The results showed that the proposed method could effectively record and measure the full crack propagation process, and the dynamic crack width variation for each pixel along the skeleton line and the crack length variation could be automatically calculated. Furthermore, the cracking initiation, propagation, coalescence, acceleration and ultimate failure can be effectively identified with the cracking growth speed variations and the crack number evolution tree. Verification experiment showed that the dynamic crack width curves measured with the proposed method were in good agreement with those obtained via the DIC method, with the root mean square error (RMSE) less than 0.1 mm and the coefficient of determination (R2) greater than 0.96. The proposed method in this paper provides a new way of cracking speed measurement that is crucial for structural crack safety valuation and technical condition assessment.