Crack Detection in Pavement Images Based on a Self-​Adaptive Niche Algorithm

Journal of Applied Science and Engineering, Vol. 25, No 3, Page 513-526
513
Crack Detection in Pavement Images Based on a Self-​Adaptive Niche Algorithm

This study focuses on optical image pavement damage detection instead of articial detection in pavement
maintenance. Based on the characteristics of cracks and combined with the niche theory, it proposes a dynamic
adaptive curve extraction algorithm. First, we construct the matrix space, map the original pavement image
data to the target space, process the data in target space using the multi-trough algorithm, then connect the
extreme gray value points between two adjacent scanning rows with lines, compare the average gray value of
the line with the average gray value of this area, and judge the possibility of cracks according to curve extension
characteristics. The method considers oil, water stain, irregular concave spot, and other kinds of image noise on
the pavement surface. It has good adaptability, and experimental results show that the algorithm is effective.
Keywords: pavement, crack detection, image processing, niche thought