Title :
Automatic classification and recognition of defects in an eddy current non destructive testing
Author :
Simard, P. ; Gaillard, P.
Author_Institution :
Université de Technologie de Compiègne, Compiegne Cedex, France
Abstract :
The main object of this paper is to propose a method for an automatic classification of defects in an eddy current non destructive testing. This method is based on a few simple hypothesis on the shape of the defect and the reaction of the probe which permits to extract information sufficient enough for classification. The analysis of the "image" of the defect which is obtained by the scanning of the tested piece involves two steps. First, an analysis of the orientation and the shape of the defect and then the analysis of the relief of this image, which determines the position of significant features. The method has been tested on a given sample of population with a 80% rate of success.
Keywords :
Automatic testing; Data mining; Eddy current testing; Eddy currents; Image analysis; Particle measurements; Position measurement; Probes; Shape; Signal processing;
Conference_Titel :
Acoustics, Speech, and Signal Processing, IEEE International Conference on ICASSP '86.
DOI :
10.1109/ICASSP.1986.1169220