Crack Width Detection of the Concrete Surfaced Based on Images

In order to detect the crack width of the concrete surface, we propose an algorithm of local rotation of images. Firstly, we calculate the normal direction of the given point in the crack to get the rotation angel. Secondly, we take a sub-image centering in the point, and rotate the sub-image around the point to make the crack in the sub-image changing to be vertical. Finally, the crack width is defined by the number of pixels on the line where the point is. By this algorithm, we can calculate the width of every cross-section on the entire crack, and the automation of the crack width detection will be come true. Based on this, we can also derive the crack length through the number of pixels on the crack (area) and the crack width. Theoretical analysis and numerical experiments show that this algorithm has high precision and good stability.