Abstract: Due to the brightness similarity and linear consistency between surface cracks and inherent lining seams in some tunnel concrete lining,lining seams are prone to sugaring,hollowing,localized rock fall and water leakage.The existing crack detection algorithms cannot accurately extract single cracks.This paper proposed a lining seam elimination algorithm based on the extraction of line feature unit in tunnel concrete lining.Firstly,on the basis of the clustering characteristics detection which is similar to cracks,the significant linear characteristics have been detected by progressive probabilistic hough ransform(PPHT).Secondly,the minimum line feature processing unit of lining seam,which is called Unit-Line,has been extracted by the search calculation of pixel extension.Finally,according to the marking information of the Unit-Line characteristics and the Unit-Line characteristics in fixed area,a part of the lining seams can be removed,and the real surface cracks in tunnel concrete lining can be obtained by the percolation de-noising.The experimental results show that the proposed surface lining seam elimination algorithm based on the extraction of line feature unit,which has strong robustness,fills the gap in the existing technology of tunnel concrete lining surface cracks detection,and the interference of similar linear characteristics to the single real cracks detection can be removed accurately,quickly and effectively.

Key words: Lining seam elimination,PPHT,Unit-Line,Percolation de-noising,Crack detection