Water Journal : Water Journal May 2012
pipeline cleaning & maintenance refereed paper technical features 62 MAY 2012 water inspection based on segmentation of colour images by SVM has been described. Analysis of the principal segmented region by mathematical morphology enables the flow line region, pipe joints and adjoining defects to be detected. While manifesting some false positives or false negatives in corrosion detection, the system had high accuracy in the detection of flow line regions, pipe joints and pipe connections on the data on which it was tested. A technique for processing pipe image collections by growing self-organising maps to filter out good regions of pipe has also been described. By means of this processing more than 80% of pipe image data may be filtered out, leaving a much smaller amount of data to be processed either manually or by further automated systems. Further research will be directed to the problem of pipe connection detection, and then further investigation into the detection of pipe defects on a large scale (such as pipe deformation) or a small scale (such as holes or cracks). Acknowledgements Work described in this paper has been funded by the CSIRO Water for a Healthy Country Flagship. The authors thank Yuhang Zhang and Richard Hartley of ANU; Hiran Ganegedara, Damminda Alahakoon and Andrew Paplinski of Monash University and Karsten Müller and Thomas Deserno of Aachen University for collaboration on this work; Mike Rahilly for implementing a graphical user interface; and Paul Davis and David Marlow for very helpful discussions. Authors Dr John Mashford (e-mail: John.Mashford@ csiro.au) is a Senior Research Scientist with the CSIRO Division of Land and Water, PO Box 56, Highett, Victoria 3190, Australia. Dr Donavan Marney (email: Donavan.Marney@csiro.au) is a Research Scientist and Stream Leader of Intelligent Networks at CSIRO. Prof Stewart Burn (email: Stewart. Burn@csiro.au) is a Senior Principal Research Scientist at CSIRO's Land and Water Division and an Adjunct Professor at Victoria University. References Alahakoon D, Halgamuge S & Srinivasan B (2000): Dynamic self-organising maps with controlled growth for knowledge discovery, Neural Networks, IEEE Transactions on, Vol 11, No 3, pp 601--614. Cheng HD, Jiang XH, Sun Y & Wang J (2001): Color image segmentation: advances and prospects, Pattern Recognition 34, pp 2259--2281. Ganegedara H, Alahakoon D, Mashford J, Paplinski A, Müller K & Deserno T (2012): "Self-organising map-based region of interest labelling for automated defect identification in large sewer pipe image collections", accepted, to appear in Proc. 2012 International Joint Conference on Neural Networks. Kirkham R, Kearney PD, Rogers KJ & Mashford J (2000): PIRAT -- A system for quantitative sewer pipe assessment, The International Journal of Robotics Research, Vol 19, No.11, pp 1033--1053. Kuntze H, Haffner H, Selig M, Schmidt D, Janotta K & Loh M (1994): Development of a flexible utilisable robot for intelligent sensor-based sewer inspection, Proc. of the 4th International Conference on Pipeline Construction, Hamburg, Germany, 367-374 (1994). Mashford JS (1995): A neural network image classification system for automatic inspection, Proc. of the 1995 IEEE International Conference on Neural Networks, Perth, Australia. Mashford J, Davis P & Rahilly M (2007): Pixel-based colour image segmentation using support vector machine for automatic pipe inspection, Lecture Notes in Artificial Intelligence 4830, Springer- Verlag, Berlin. Mashford J, Rahilly M, Davis P & Burn S (2010): "A Morphological Approach to Pipe Image Interpretation based on Segmentation by Support Vector Machine", Automation in Construction, Vol 19(7), pp 875--883. Müller K & Fischer B (2007): Objective condition assessment of sewer systems, LESAM 2007 -- 2nd Leading Edge Conference on Strategic Assessment Management, Lisbon, Portugal. Sinha SK & Fieguth PW (2006): Segmentation of buried concrete pipe images, Automation in Construction, Vol 15, pp 47--57. Zhang Y, Hartley R, Mashford J, Wang L & Burn S (2011): "Pipeline reconstruction from fisheye images", Proc. 19th International Conference on Computer Graphics, Visualization and Computer Vision 2011, Plzen, Czech Republic. Figure 10. Overview of system for processing sewer image collections by growing self-organising map.
Water Journal July 2012
Water Journal April 2012