Image-based crack detection approaches: a comprehensive survey

Automatic crack detection is a challenging task that has been researched for decades due to the complex civil structures. Cracks on any structure are early signs of the deterioration of the object’s surface. Therefore, detection and regular maintenance of cracks are necessary tasks as the propagation of cracks results in severe damage. Manual inspection is based on the expert’s previous knowledge, and it can only be done in reachable human areas. On the other hand, autonomous detection of cracks by using image-based techniques may reduce human errors, less time-consuming, and more economical than human-based inspection for real-time crack detection. Since movable cameras can capture images for non-reachable areas, several techniques are available for crack detection. Several techniques are available for crack detection; however, image-based crack detection techniques have been analyzed in this survey. A detailed study is carried out to define the research problems and advancements in this area. This article analyses the pure image processing techniques and learning-based techniques based on the objectives, the methods, level of efficiency, level of errors, and type of crack image dataset. Besides the applications, limitations and other factors are explained for each technique. Moreover, the presented analysis shows the multiple problems related to cracks that could help the researcher perform further research.