Performance of Image-Based Crack Detection Systems in Concrete Structures

Abstract

The traditional methods for calculating the width of the cracks in concrete structures are mainly based on the manual and non-systematic collection of information, and also depend on personal justifications and judgment. Due to the fact that these approaches are time-consuming and always there are some human errors inevitably, in recent years more attention is paid to the new methods for detection and monitoring of cracks. One of the most important new approaches is the application of image-based techniques. These schemes use field images and photos provided by the camera to determine specific parameters, such as damage occurrence, location, severity, length of cracks, width and depth of cracks. Moreover, tracking the crack propagation over time using a set of timed photos is among the design purposes of these methods. Image processing, and targeting are two common methods which have their own pros and cons. Results showed that the image processing approach detects some surface noises as cracks which is most challenging error in this method. On the other hand, targeting approach has shown weakness in determining the exact location of cracks. These limitations have pushed researchers to innovate more modern techniques such as Digital Image Correlation (DIC) and mathematical tools like Wavelet transform (WT) to eliminate these errors.