Image analysis for crack detection

Evaluation of cracks using eddy currents (ECT) is one of the most used methods for non destructive testing (NDT). However, this technique still needs to be improved. One of the problems is related to the fact that eddy currents tests are so sensitive that a reproducible background noise signal is always present mitigating the informative data. This paper presents new developments in the area of signal processing that promise significant improvements to an accurate data interpretation. The steps carried out in the processing algorithm are illustrated using the experimental data acquired in the detection of a crack in an aluminum plate.