Home
Home
16th WCNDT 2004 - World Conference on NDT
CD-ROM Proceedings, Internet Version of ~600 Papers
Aug 30 - Sep 3, 2004 - Montreal, Canada
START
Home

First 1st    previous prevWCNDT 2004 - Abstractsnext next

SESSION: SIGNAL PROCESSING
ABSTRACT:
AUTOMATIC CLASSIFICATION OF DEFECTS IN TIME-OF-FLIGHT-DIFFRACTION 
O. Zahran, W. Al-Nuaimy
Department of Electrical Engineering & Electronics, The University of Liverpool, Brownlow Hill, 
Liverpool, United Kingdom

Ultrasonic Time-Of-Flight Diffraction (TOFD) is a recent innovation that has proved highly effective for the 
inspection of steel plates and tubular pipelines and has started to take its way to replace the other ultrasonic 
testing techniques.

TOFD technique has a lot of advantages which make it the preferable technique in material testing. This 
technique gives accurate sizing, positioning and characterising of weld and other defects with a high 
probability of detection. 

It is anticipated that coupled with the necessary processing algorithms, TOFD can be used for a 
comprehensive automatic inspection of the weld areas with satisfactory levels of accuracy and reliability.

Currently most of the TOFD data interpretation is done manually, requiring operator skill, experience and 
most significantly time. In light of the industrial pressure, the recent trend is to partially or fully automate 
the inspection and data interpretation process, which could potentially improve the inspection and 
interpretation procedures by adding and element of robustness and consistency by utilising computational 
tools that are better suited to discriminating between subtle variations in visual and spectral properties of the 
data; furthermore, save money, effort and time.

Advanced signal and image processing techniques have been developed to characterise TOFD signals and 
extract distinguishable features to be used in defect detection and classification. Several features have been 
investigated and selected, which prove to produce a good discrimination between defect and non-defect and 
also between different defect types. 

Combining these features with a multi-layer perception artificial neural network allows the TOFD images 
segmentation and classification of detected defects in a fully automatic and un-supervised manner. 
Full-Text HTML-txtQuick PDF Preview
Full-Text PDF (KB)pdfPDF 353
Full-Text HTML:
OPTION (MB):
MAIN AUTHOR:Osama Zahran, The University Of Liverpool, United Kingdom
Paper CODE: 439

© NDT.net