Cracking is one of the major distresses impacting pavement quality, serviceability, and lifespan. Thus, accurate, precise, and complete cracking detection is important in the maintenance, performance evaluation, structure, and material design of pavements. Given that the results of pavement crack image recognition tend to contain noises and intermittent crack segments, an automatic crack detection algorithm based on the connectivity checking of pixels and crack block levels was proposed. First, a pavement image was enhanced on the basis of a self-adaptive grayscale stretch. The image was then segmented into background and foreground (potential cracks) on the basis of self-adaptive OTSU segmentation and 8-direction Sobel gradients. The potential crack image was denoised through connectivity checking. Finally, crack blocks were detected and optimally connected to form the final crack image. Examples show that the results of the proposed algorithm maintain enhanced integrity and continuity for improving connectivity at both the pixel and block levels. Performance tests were also conducted on 10 pavement images () for global OTSU segmentation, 8-direction Sobel detection, Canny detection, and the proposed algorithm. The proposed algorithm achieved both the highest precision (86.60%) and recall (90.68%), which resulted in the best score ().
Pavement Crack Detection Algorithm Based on Bi-Layer Connectivity Checking
2014-11-14
102014-01-01 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Unbekannt
Internal Crack Detection Algorithm of High Speed Multi Layer Asphalt Pavement Engineering
Springer Verlag | 2022
|