科研之友
微信
新浪微博
Facebook
分享链接
摘要
This paper develops a fully automatic algorithm to detect steel rebar in concrete bridge decks from ground penetrating radar (GPR) data. The surveyed GPR data from reinforced concrete (RC) structure carries significant amount of information, such as rebar locations, dielectric constant of concrete, corrosion conditions of steel rebars, and etc. However, current methodologies to identify rebar hyperbolas from GPR data are either manual or interactive, which is time-consuming. An unsupervised, efficient, and robust pattern recognition algorithm is proposed. In the algorithm, a template of a rebar hyperbola is selected first from the data, and is updated automatically to achieve better performance of rebar detection during the data processing. A rebar hyperbola in the GPR data is detected from the local minimum points on the map of sum of square difference (SSD) with an adaptive threshold value. An output file is created and saved, containing the amplitudes of rebar reflection and the geometrical locations of the detected rebars. The proposed algorithm is tested with ground-coupled GPR data collected from several decommissioned RC slabs, which are saw cut from a highway bridge. The overall percentage of accurate rebar hyperbola detection is 88.0%. Finally, the reflection amplitudes extracted from the rebar hyperbolas are used to create a corrosion indicator map, which shows good agreement with the results of half-cell potential maps.
- 出版日期2018-11
- 单位东北大学
