Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 108
PROCEEDINGS OF THE FIFTEENTH INTERNATIONAL CONFERENCE ON CIVIL, STRUCTURAL AND ENVIRONMENTAL ENGINEERING COMPUTING
Edited by: J. Kruis, Y. Tsompanakis and B.H.V. Topping
Paper 283

Patch Defects in Images using Support Vector Machines

G. Hadjidemetriou1, M.M. Serrano2, P.A. Vela2 and S. Christodoulou1

1Department of Civil and Environmental Engineering, University of Cyprus, Nicosia, Cyprus
2Department of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, United States of America

Full Bibliographic Reference for this paper
G. Hadjidemetriou, M.M. Serrano, P.A. Vela, S. Christodoulou, "Patch Defects in Images using Support Vector Machines", in J. Kruis, Y. Tsompanakis, B.H.V. Topping, (Editors), "Proceedings of the Fifteenth International Conference on Civil, Structural and Environmental Engineering Computing", Civil-Comp Press, Stirlingshire, UK, Paper 283, 2015. doi:10.4203/ccp.108.283
Keywords: pavement condition assessment, patch detection, computer vision, image processing, support vector machine, MATLAB.

Summary
Condition assessment of roadway transport networks evaluates the vulnerability of road segments and by extension the future reliability of such networks. Traditional road pavement condition assessment involves manual effort, which requires training, is costly and time-consuming given the quantity of road needing assessment, and poses a risk to the safety of inspectors involved. Automated roadway assessment methods using computer vision can overcome these limitations. In this paper, the detection of patch type distresses in images, taken from video cameras positioned on vehicles is presented. This method is based on support vector machine (SVM) classification applied to the image data. The process first divides a test image into square blocks, then generates a feature vector from the block to be used in the SVM. Two feature vector types are tested: histogram and texture using the discrete cosine transform. The result is an output with each image block being classified as "patch" or "no-patch". The recognition results of the case study from a real-life urban network are promising, showing a detection accuracy of about 81%, based on a MatlabTM implementation

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £75 +P&P)