A novel tunnel-lining crack recognition system based on digital image technology

•An automated visual system for recognizing tunnel cracks is proposed.•Prior noise judgment achieves differentiated noise filtering.•The system combines adaptive partitioning, double threshold and edge detection. Structural health monitoring (SHM) combined with digital image technology has been wide...

Full description

Saved in:
Bibliographic Details
Published in:Tunnelling and underground space technology 2021-02, Vol.108, p.103724, Article 103724
Main Authors: Lei, Mingfeng, Liu, Linghui, Shi, Chenghua, Tan, Yuan, Lin, Yuexiang, Wang, Weidong
Format: Article
Language:English
Subjects:
Adaptive filters
Algorithms
Calibration
Crack propagation
Digital imaging
Edge detection
Factor analysis
Feature extraction
Forecasting
Image enhancement
Image processing systems
Imaging techniques
Inspection
Object recognition
Recognition system
Reliability analysis
Reliability aspects
Stability analysis
Structural health monitoring
Tunnel crack
Tunnel linings
Tunnels
Underground construction
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•An automated visual system for recognizing tunnel cracks is proposed.•Prior noise judgment achieves differentiated noise filtering.•The system combines adaptive partitioning, double threshold and edge detection. Structural health monitoring (SHM) combined with digital image technology has been widely applied to infrastructure operation management. However, the linear illumination of the tunnel and the various lining diseases limit the quality of lining-crack recognition. In this article, a novel tunnel-lining crack recognition system is established. The system involves three main procedures: image preprocessing and enhancement, feature extraction, and crack characterization. To meet tunnel environmental conditions, mature image enhancement and morphological algorithms are packaged into the system; meanwhile, this paper proposes differentiated noise filtering and an improved segmenting method combining adaptive partitioning, edge detection and threshold method to improve the recognition accuracy. A self-regulating calibration method that uses parallel projection is also applied to crack characterization, achieving real-time size calibration. The results of experiments to compare the effects of the proposed system and field application tests confirm the stability and reliability of the system. A further deviation factor analysis provides reasonable suggestions for system improvement.
ISSN:0886-7798
1878-4364
DOI:10.1016/j.tust.2020.103724