Eddy current testing has the advantage of non-contact and
fast test method over other non-destructive testing methods.
However, one of the disadvantages of eddy current testing is
that it tends to generate large noise generated by variations
of many factors such as probe lift-off and electromagnetic
characteristics of the test material. For the particular case of
welded zone testing the conventional eddy current
method [1] performance is strongly affected by shape
changes and variable electromagnetic characteristics on the
tested regions. In order to overpass these problems several
eddy current probe architectures are reported in the
literature [2] and are normally based on excitation and
detection coil sets [3]. Latest works on flaw and crack
detection in conductive plates underline the capabilities of
eddy current probes (ECP) based on single or multiple giant
magneto-resistance (GMR) sensors [4]. The use of GMRs is
a good tested choice for the present application.
When a GMR is used the obtained images represent one
component of the magnetic induction field along the sensing
axis. With these images it is possible to detect the presence of flaws. The estimation of the geometrical characteristics of
the flaws is carried out using the amplitude and the phase
evolution profile for a tested area [5]. Normalization,
filtering, image segmentation, feature extraction and
classification techniques [6] are used to evaluate the specific
characteristics of those detected flaws.
This paper continues previous works on non-destructive
testing of aluminum plates [7,8]. However, a particular case
of flaw detection in an aluminum plate welded zone is
considered. New elements concerning the specific eddy
current test (ECT) probe architectures including coils and
GMR sensors are an important part of this work. Referring
to the image processing that is particularly associated with
the flaw detection in the welded zone, a practical approach
on optimal image filtering for a better detection of the flaws
is included in the work.