Abstract:

The layout image of a lining surface is essential for structural safety inspections of operational tunnels. In order to identify information from a layout image, this study employs supervised learning image interpretation technology to develop an anomaly detection method and then uses the Hough transform to characterize crack distribution patterns. Application results show that the proposed technology can efficiently and precisely identify cracks from lining surface image, and can give distinct numbers corresponding to different special distribution patterns of cracks. The proposed technology is also applicable to cracks that have similar special distribution pattern but have different level of development.

Key words: tunnel lining, crack pattern, image interpreting, characterizing, digitalizing