Please wait a minute...
Frontiers of Environmental Science & Engineering

ISSN 2095-2201

ISSN 2095-221X(Online)

CN 10-1013/X

Postal Subscription Code 80-973

2018 Impact Factor: 3.883

Front. Environ. Sci. Eng.    2019, Vol. 13 Issue (2) : 17    https://doi.org/10.1007/s11783-019-1102-y
RESEARCH ARTICLE
Diagnosis of sewer pipe defects on image recognition of multi-features and support vector machine in a southern Chinese city
Xiangyang Ye1, Jian’e Zuo1(), Ruohan Li1,2, Yajiao Wang1, Lili Gan1,3, Zhonghan Yu1, Xiaoqing Hu1,2
1. State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, China
2. Thunip Corp., Ltd., 27F/Section C, S&T Building, Tsinghua S&T Park, Beijing 100084, China
3. China Water Environment Group. 10F, Campbell Building, 89th Jinbao Street, Beijing 101101, China
 Download: PDF(2687 KB)   HTML
 Export: BibTeX | EndNote | Reference Manager | ProCite | RefWorks
Abstract

An image-recognition-based diagnosis system of pipe defect types was established.

1043 practical pipe images were gathered by CCTV robot in a southern Chinese city.

The overall accuracy of the system is 84% and the highest accuracy is 99.3%.

The accuracy shows positive correlation to the number of training samples.

Closed circuit television (CCTV) systems are widely used to inspect sewer pipe conditions. During the diagnosis process, the manual diagnosis of defects is time consuming, labor intensive and error prone. To assist inspectors in diagnosing sewer pipe defects on CCTV inspection images, this paper presents an image recognition algorithm that applies features extraction and machine learning approaches. An algorithm of image recognition techniques, including Hu invariant moment, texture features, lateral Fourier transform and Daubechies (DBn) wavelet transform, was used to describe the features of defects, and support vector machines were used to classify sewer pipe defects. According to the inspection results, seven defects were defined; the diagnostic system was applied to a sewer pipe system in a southern city of China, and 28,760 m of sewer pipes were inspected. The results revealed that the classification accuracies of the different defects ranged from 51.6% to 99.3%. The overall accuracy reached 84.1%. The diagnosing accuracy depended on the number of the training samples, and four fitting curves were applied to fit the data. According to this paper, the logarithmic fitting curve presents the highest coefficient of determination of 0.882, and more than 200 images need to be used for training samples to guarantee the accuracy higher than 85%.

Keywords Sewer pipe defects      Defect diagnosing      Image recognition      Multi-features extraction      Support vector machine     
Corresponding Author(s): Jian’e Zuo   
Issue Date: 25 January 2019
 Cite this article:   
Xiangyang Ye,Jian’e Zuo,Ruohan Li, et al. Diagnosis of sewer pipe defects on image recognition of multi-features and support vector machine in a southern Chinese city[J]. Front. Environ. Sci. Eng., 2019, 13(2): 17.
 URL:  
https://academic.hep.com.cn/fese/EN/10.1007/s11783-019-1102-y
https://academic.hep.com.cn/fese/EN/Y2019/V13/I2/17
Fig.1  The sewer pipe map of the southern city in China.
Fig.2  The basic information of the inspected sewer pipes. (a) The distribution of ages; (b) The distribution of materials; (c) The distribution of diameters.
Fig.3  Typical pipe defect images.
Fig.4  Optimal classification hyperplane.
Fig.5  Schematic overview of the diagnosis system.
Image Classified vector Diagnosed result Real defect
(6,2,2,2,5,4,0) Deformation Deformation
(3,0,2,6,5,4,1) Attached deposits Attached deposits
(5,0,3,2,6,4,1) Settled deposits Attached deposits
Tab.1  The classified vectors and diagnosed results
Fig.6  Original image and processed images (a) Original image. (b) Greyscale image. (c) Lateral Fourier analysis image. (d) DB4 wavelet transform image.
Hu invariants Value Texture feature Value
M1 0.3224 Average gray level 134.6
M2 0.0036 Average contrast 44.93
M3 5.322 × 10-5 Smoothness 0.0301
M4 2.460 × 10-5 The third moment 1.281
M5 -3.707 × 10-10 Uniformity 0.0086
M6 -2.342 × 10-7 Entropy 7.214
M7 8.093 × 10-10
Tab.2  The Hu invariants and texture feature values of the example image
Fig.7  The feature value distributions of 1043 defect images. (a) The Hu invariant moments value distributions; (b) the texture feature value distributions; (c) The spectrum value distributions.
Fig.8  Distributions of every two features among the three selected features. (a) Features of average lateral spectrum vs. Hu invariant moment-2.; (b) Features of Hu invariant moment-2 vs. smoothness; (c) Features of smoothness vs. average lateral spectrum.
Defect type Total number of images Number of correctly diagnosed images Diagnosis accuracy (%)
Deformation 202 162 80.2
Collapse 31 16 51.6
Infiltration 105 83 79.1
Attached deposits 83 63 75.9
Settled deposits 403 400 99.3
Displaced joint 169 126 74.6
Joint damage 52 29 55.8
Overall 1045 879 84.1
Tab.3  Diagnosis result and classification accuracy of the SVM classifier
Defect type Average number of training samples Classification accuracy (%)
Deformation 181.8 80.2
Collapse 27.9 51.6
Infiltration 94.5 79.1
Attached deposits 74.7 75.9
Settled deposits 362.7 99.3
Displaced joint 152.1 74.6
Joint damage 46.8 55.8
Tab.4  Classification accuracy of sewer pipe defects and the number of training samples
Fig.9  The fitting curves of training sample number and classification accuracy.
1 Canadian Standards Association ( 2010). Canadian Standards Association Technical Guide- PLUS 4012: Visual Inspection of Sewer Pipe. Canadian Standards Association, Toronto, ON, Canada
2 L MDang, S I Hassan, S Im, IMehmood, HMoon (2018). Utilizing text recognition for the defects extraction in sewers CCTV inspection videos. Computers in Industry, 99: 99–109
https://doi.org/10.1016/j.compind.2018.03.020
3 JDirksen, F H L R Clemens, H Korving, FCherqui, PLe Gauffre, TErtl, H Plihal, KMüller, C T MSnaterse (2013). The consistency of visual sewer inspection data. Structure and Infrastructure Engineering, 9(3): 214–228
https://doi.org/10.1080/15732479.2010.541265
4 L LGan, J E Zuo, Y J Wang, T S Low, K J Wang (2014). Comprehensive health condition assessment on partial sewers in a southern Chinese city based on fuzzy mathematic methods. Frontiers of Environmental Science & Engineering, 8(1): 144–150
https://doi.org/10.1007/s11783-013-0554-8
5 WGuo, L Soibelman, J HGarrett Jr (2009). Automated defect detection for sewer pipeline inspection and condition assessment. Automation in Construction, 18(5): 587–596
https://doi.org/10.1016/j.autcon.2008.12.003
6 M RHalfway, J Hengmeechai (2014). Automated defect detection in sewer closed circuit television images using histograms of oriented gradients and support vector machine. Automation in Construction, 38: 1–13
https://doi.org/10.1016/j.autcon.2013.10.012
7 M RHalfawy, J Hengmeechai (2015). Integrated vision-based system for automated defect detection in sewer closed circuit television inspection videos. Journal of Computing in Civil Engineering, 29(1): 04014024
https://doi.org/10.1061/(ASCE)CP.1943-5487.0000312
8 AHawari, M Alamin, FAlkadour, MElmasry, TZayed (2018). Automated defect detection tool for closed circuit television (CCTV) inspected sewer pipelines. Automation in Construction, 89: 99–109
https://doi.org/10.1016/j.autcon.2018.01.004
9 MHu (1962). Visual-pattern recognition by moment invariants. I.R.E. Transactions on Information Theory, 8(2): 179–187
https://doi.org/10.1109/TIT.1962.1057692
10 SIyer, S K Sinha (2005). A robust approach for automatic detection and segmentation of cracks in underground pipeline images. Image and Vision Computing, 23(10): 921–933
https://doi.org/10.1016/j.imavis.2005.05.017
11 S SKumar, D M Abraham, M R Jahanshahi, T Iseley, JStarr (2018). Automated defect classification in sewer closed circuit television inspections using deep convolutional neural networks. Automation in Construction, 91: 273–283
https://doi.org/10.1016/j.autcon.2018.03.028
12 XLiang (2006). Support Vector Machine-Based Image Classification Research. Dissertation for the Master Degree. Shanghai: Tongji University (in Chinese)
13 JMashford, P Davis, M (Rahilly 2007).Pixel-based color image segmentation using support vector machine for automatic pipe inspection. In: M A Orgun, J Thornton, eds., Proc. of the 20th Australian Joint Conference on Artificial Intelligence. Heidelberg: Springer-Verlag, : 739–743
14 JMashford, M Rahilly, PDavis, SBurn (2010). A morphological approach to pipe image interpretation based on segmentation by support vector machine. Automation in Construction, 19(7): 875–883
https://doi.org/10.1016/j.autcon.2010.06.001
15 National Bureau of Statistics of China. China Statistical Yearbook 1998–2013. Beijing: China Statistic Press, 1998–2013 (in Chinese)
16 OPanasiuk, A Hedström, JMarsalek, R MAshley, MViklander (2015). Contamination of stormwater by wastewater: A review of detection methods. Journal of Environmental Management, 152: 241–250
https://doi.org/10.1016/j.jenvman.2015.01.050 pmid: 25662485
17 TShehab, O Moselhi (2005). Automated detection and classification of infiltration in sewer pipes. Journal of Infrastructure Systems, 11(3): 165–171
https://doi.org/10.1061/(ASCE)1076-0342(2005)11:3(165)
18 T CSu, M D Yang (2014). Application of morphological segmentation to leaking defect detection in sewer pipelines. Sensors (Basel), 14(5): 8686–8704
https://doi.org/10.3390/s140508686 pmid: 24841247
19 Q ATran, Q L Zhang, X Li (2003). Reduce the number of support vectors by using clustering techniques. In: Proceedings of the 2nd international conference on machine learning and cybernetics, Xi’an, China, 1245–1248
20 M DYang, T C Su (2008). Automated diagnosis of sewer pipe defects based on machine learning approaches. Expert Systems with Applications, 35(3): 1327–1337
https://doi.org/10.1016/j.eswa.2007.08.013
21 M DYang, Y F Yang, T C Su, K S Huang (2014). An efficient fitness function in genetic algorithm classifier for landuse recognition on satellite images. The Scientific World J,,
https://doi.org/10.1155/2014/264512
22 JZhang (2013). Research and Application of Diagnosis Technologies for Crop Pests based on Image Recognition. Dissertation for the Doctoral Degree. Hefei: University of Science and Technology of China (in Chinese)
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed