Abstract:

To realize the automatic detection of cracks during the intelligent patrol in metro tunnels, this study presents a texture feature analysis-based crack detection scheme for metro tunnels. In the proposed scheme, several pre-processing operations such as erosion, contrast stretch and weighted neighborhood are firstly adopted to improve the quality of original scanned images. Then, the images are partitioned and the improved maximum between-cluster variance method is utilized for the partitioned areas so as to separate the cracks from the background images. Next, the texture feature analysis is leveraged to filter out incorrect crack information from the images. Afterwards, the images are thinned so as to obtain the corresponding skeleton images, and finally the automatic detection of cracks for metro tunnels can be realized. Furthermore, the experimental results demonstrate that the proposed scheme can efficiently and accurately detect the crack information for metro tunnels as well as automatically create labels for the crack areas, and the detection effect of the proposed scheme is significantly better than the existing schemes.

Key words: crack detection, metro tunnel, texture feature analysis, automatic detection, weighted neighborhood filter, OTSU, image thinning, direction chain code