Int. J. Computer Aided Engineering and Technology, Vol. 1, No. 2, 2009 225 Copyright © 2009 Inderscience Enterprises Ltd. Automatic quality control of seam puckers based on shadow detection Ioannis G. Mariolis* Department of Electrical Engineering and Computer Technology, University of Patras, 26500, Patras, Greece Fax: +30 2610 996 476 E-mail: mariolis@george.wcl2.ee.upatras.gr *Corresponding author Evangelos S. Dermatas Department of Electrical Engineering and Computer Technology, University of Patras, 26500, Patras, Greece Fax: +30 2610 996 189 E-mail: dermatas@george.wcl2.ee.upatras.gr Abstract: A novel method performing automatic seam quality control based on visual information is introduced. More specifically, oblique illumination is applied and greyscale images of seam specimens are acquired. The amount of shadowing present in each image is being estimated through dynamic thresholding and first order statistics and is used by a simple linear classifier to estimate the seam quality by means of ordinary least squares (OLS) regression. The proposed method has been evaluated in the case of 112 seam specimens classified by human experts into five discrete quality grades and a correct classification rate of 84.82% has been produced exceeding the experts’ average performance. Moreover, in case one grade disagreement is acceptable, the classification results become error free. Keywords: seam quality; seam puckering; machine vision; shadow detection; regression. Reference to this paper should be made as follows: Mariolis, I.G. and Dermatas, E.S. (2009) ‘Automatic quality control of seam puckers based on shadow detection’, Int. J. Computer Aided Engineering and Technology, Vol. 1, No. 2, pp.225–238. Biographical notes: Ioannis G. Mariolis was born in 1979 in New Jersey, USA. He received his Diploma degree in Engineering from the Department of Electrical Engineering and Computer Technology of the University of Patras, Patras, Greece in 2002 and in the same year was accepted as a PhD Student of the same department. His research interests include machine vision, visual servoing, pattern recognition and signal processing. Evangelos S. Dermatas is an Assistant Professor in the Department of Electrical and Computer Engineering of the University of Patras, Patras, Hellas. He received his Diploma and PhD degrees from the Department of Electrical Engineering of the University of Patras, Patras, Greece in 1985 and 1991 respectively. His research interests include statistical signal processing, pattern recognition, computer security and information extraction.