| description abstract | Automatic pavement cracking detection is of great practical value for pavement maintenance, pavement performance evaluation and prediction, and material and structure design. However, detecting pavement cracks rapidly, precisely, completely, and robustly remains a challenge. Thus, literature review on automatic pavement track detection was conducted, which included pre - processing methods aiming at image enhancement and de - noising, space - domain recognition algorithms based on thresholding, edge detection and seed growing, frequency - domain recognition algorithms, such as wavelet transform, and supervised learning methods. Shortcomings of these crack detection algorithms were summarized as follow: (1) illumination and oils tend to affect algorithm performance; (2) crack maps have poor continuity; (3) processing speed and recognition precision are not satisfying. Research prospects were also proposed as references to improve crack recognition algorithms, including (1) removing influences of texture and noises by combining boundary and area features, (2) designing optimization - based - recognition algorithms that consider local and global features, and (3) detecting pavement cracks basted on 3D images. | |
