Kuwait Journal of Science

Previous Issues

Advance Search
Year : From To Vol
Issue Discipline:

Volume :36 Issue : 2 2009      Add To Cart                                                                    Download

A real time image processing algorithm for cracks highlighting in weld x-ray video


Department of Electrical Engineering, Faculty of Engineering, University of Isfahan.

Hezarjarib Avenue, Isfahan, 81746-73441, Iran.

Email: p_moallem@eng.ui.ac.ir

**Department of Mechanics, Mohajer Technical Institute

Hezarjarib Avenue, Isfahan, Iran.

Email: a_moallem@yahoo.com




Welding radiography is a common method in the quality control of welds. Good human interpreting of radiography images is time consuming. Therefore using intelligent methods in radiography interpreting can improve the quality of interpretations as well as reducing interpreting time. In X-Ray weld inspection systems with video output, it is necessary to detect defects very fast. As an assistant for weld images interpreters, this paper introduces an image processing algorithm with low computational cost for highlighting of cracks that may be lost in a video based X-Ray weld inspection system. The interpreter can then detect true cracks in the highlighted candidate positions faster and more accurate.


To alleviate the noise effects in the digital input images, a 7\times7 Gaussian filter with variance of 1.5 is applied. Then the vertical projections of the filtered images are used to accurately detect weld region of the input images. The detected weld region is segmented to rectangular regions. The horizontal projection of the rectangular regions is analyzed to highlight the transverse cracks, while the vertical projection of rectangular regions is used to highlight longitude cracks. Some parameters of the proposed algorithm can be adjusted by user to obtain required accuracy.


The implementation results of the proposed algorithm for the real radiography weld images show the power of the algorithm to highlight transverse and longitude cracks. To evaluate the performance of the proposed image processing algorithm for video images in real time application, a visual C++ programming implementation of the algorithm is developed. The execution time of the developed application for video sequences in a PC with a 2.0 GHz Core Dou CPU is less than 40 milliseconds per image. Therefore it is suitable for real time cracks highlighting of online welding X-Ray inspection systems with video output.


Keywords: Cracks Highlighting, Gaussian Filter, Projections, Radiography Welding Images.

Kuwait Journal of Science
Journal of Law

You are Visitor No.


Journal of Law
Journal of Law
Tell your friendsJournal of Law
Journal of Law

Last Updated

Jun 19, 2012

Journal of Law
Journal of Law
Journal of Law

Please enter your email Here to receive our news

Journal of Law