Erhan AKIN, Yunus SANTUR, Mehmet KARAKÖSE

An adaptive fault diagnosis approach using pipeline implementation for railway inspection

Railway tracks must be periodically inspected. This study proposes a new approach for eliminating two majordisadvantages experienced during rail inspection applications performed via computer vision. The first is the blurringeffect on images, resulting from physical vibration during movement on the rail lines. This effect significantly reduces thehigh accuracy rate expected from anomaly inspection algorithms. The second disadvantage is the need to operate in realtime. This study presents a new three-stage computer vision method approach that eliminates both disadvantages. First,a three-stage pipeline architecture is implemented and IMU-assisted blur detection is performed on images taken fromthe left and right rail lines. Next, a convolutional neural network is used for learning. In the third test stage, anomalydetection and classification training are conducted. By performing the implementation with parallel programming ongraphic processing units, a highly accurate, cost-effective computer vision rail inspection, based on image processing andcapable of operating in real time, is successfully carried out.

PDF

___

[1] Koch C, Georgieva K, Kasireddy V, Akinci B, Fieguth P. A review on computer vision based defect detection and condition assessment of concrete and asphalt civil infrastructure. Adv Eng Inform 2015; 29: 196-210.
[2] Santur Y, Karak¨ose M, Akın E. IMU based adaptive blur removal approach using image processing for railway inspection. In: IEEE 2016 International Conference on Systems, Signals and Image Processing; 23–25 May 2016; Bratislava, Slovakia. New York, NY, USA: IEEE. pp. 1-4.
[3] Gibert X, Patel VM, Chellappa R. Deep multitask learning for railway track inspection. IEEE T Intell Transp 2017; 18:153-164.
[4] Flusser J, Suk T, Boldys J, Zitova B. Projection operators and moment invariants to image blurring. IEEE T Pattern Anal 2015; 37: 786-802.
[5] Ito A, Sankaranarayanan AC, Veeraraghavan A, Baraniuk RG. Blurburst: removing blur due to camera shake using multiple images. ACM T Graph 2014; 3: 1-15.
[6] Wang L, Hang Y, Luo S, Luo X, Jiang X. Deblurring Gaussian-blur images: a preprocessing for rail head surface defect detection. In: IEEE 2011 International Conference on Service Operations, Logistics and Informatics; 10–12 July 2011; Beijing, China. New York, NY, USA: IEEE. pp. 451-456.
[7] Li Q, Ren S. A real-time visual inspection system for discrete surface defects of rail heads. IEEE T Instrum Meas 2012; 61: 2189-2199.
[8] Nvidia. Programming Guide: CUDA Toolkit Documentation. Santa Clara, CA, USA: Nvidia, 2008.
[9] Roohi S, Hajizadeh S, N´u˜nez A, Babuska R, De Schutter B. Deep convolutional neural networks for detection of rail surface defects. In: IEEE 2016 International Joint Conference on Neural Networks; 24–29 July 2016; Vancouver, Canada. New York, NY, USA: IEEE. pp. 2584-2589.
[10] Weimer D, Scholz-Reiter B, Shpitalni M. Design of deep convolutional neural network architectures for automated feature extraction in industrial inspection. CIRP Ann-Manuf Tech 2016; 65: 417-420.
[11] Liu Z, Sun J, Wang H, Zhang G. Simple and fast rail wear measurement method based on structured light. Opt Laser Eng 2011; 49:1343-1351.
[12] Li Q, Zhong Z, Liang Z, Liang Y. Rail inspection meets big data: methods and trends. In: IEEE 2015 International Conference on Network-Based Information Systems; 2–4 September 2015; Taipei, Taiwan. New York, NY, USA: IEEE. pp. 302-308.
[13] Hackel D, Stein I, Maindorfer ML, Reiterer A. Track detection in 3D laser scanning data of railway infrastructure. In: IEEE 2015 Instrumentation and Measurement Technology Conference; 11–14 May 2015; Pisa, Italy. New York, NY, USA: IEEE. pp. 693-698.
[14] Aytekin C, Rezaeitabar Y, Dogru S, Ulusoy I. Railway fastener inspection by real-time machine vision. IEEE T Syst Man Cy-S 2015; 45: 1101-1107.
[15] Huber-M¨ork R, N¨olle M, Oberhauser A, Fischmeister E. Statistical rail surface classification based on 2D and 21/2D image analysis. In: 2010 Conference on Advanced Concepts for Intelligent Vision Systems Conference; 13–16 December 2010; Sydney, Australia. pp. 50-61.
[16] Santur Y, Karak¨ose M, Akın E. Learning based experimental approach for condition monitoring using laser cameras in railway tracks. Int J Appl Math, Electron 2016; 4: 1-5.
[17] Karak¨ose M, Aydin, Akin E. The intelligent fault diagnosis frameworks based on fuzzy integral. In: IEEE 2010 Power Electronics Electrical Drives Automation and Motion International Conference; 14–16 June 2010; Pisa, Italy. New York, NY, USA: IEEE. pp. 1634-1639.
[18] Li Y, Zuo MJ, Lin J, Liu J. Fault detection method for railway wheel flat using an adaptive multiscale morphological filter. Mech Syst Signal Pr 2017; 84: 642-658.
[19] Xsens Technologies. MTi User Manual. Enschede, the Netherlands: Xsens Technologies, 2014.
[20] Madgwick SO. An efficient orientation filter for inertial and inertial/magnetic sensor arrays. Bristol, UK: University of Bristol, 2010.
[21] Mako G-032. User Manual. Stadtroda, Germany: Allied Visions Technology, 2008.
[22] Abadi M, Barham P, Chen J, Chen Z, Davis A, Dean J, Kudlur M. TensorFlow: a system for large-scale machine learning. In: USENIX 2016 Conference on Operating Systems Design and Implementation; 2–4 November 2016; Savannah, GA, USA. Berkeley, CA, USA: USENIX Association. pp. 265-283.