Automated crack assessment and quantitative growth monitoring

Crack assessment has remained one of the high‐priority research topics for structural health monitoring. However, the current research mainly focuses on the crack assessment at some point, but pays relatively less attention to the long‐term development of cracks, which is important for structure health monitoring. In this paper, a new method based on dual‐convolutional neural network (CNN) (well over 94% accuracy), digital image processing technology and shape context is proposed, which achieves a fully automated process composed of crack detection, crack measurement, and quantitative crack growth monitoring. In crack growth monitoring, an algorithm to label each crack is put forward for the first time, which is able to reflect the sequential order of the occurrence of cracks. Therefore, each skeleton point of cracks will be assigned an ID which contains information about its identity and width in order to monitor cracks at both a global and local level. Experimental studies of a concrete member with complex cracks are utilized for the illustration and validation of the proposed methodology.