Cultural Heritage Applications: Archaeological Ceramics and Building Restoration - On Statistical Pattern Recognition in Independent Component Analysis Mixture Modelling - page 127

Chapter 6
Cultural Heritage Applications:
Archaeological Ceramics and Building
Restoration
This chapter presents two applications: classification of archaeological ceramics
and diagnosis of historic building restoration. In the first application, we consider
the ICAMM-based algorithm proposed in Chap. 3 (Mixca) to model the joint-
probability density of the features. This classifier is applied to a challenging novel
application: classification of archaeological ceramics. ICAMM by Mixca gathers
relevant characteristics that have general interest in the area of material classifi-
cation. On one hand, it can deal with arbitrary forms of the underlying probability
densities in the feature vector space as non-parametric methods can do. On the
other hand, mutual dependences among the features are modelled in a parametric
form so that ICAMM based on Mixca can achieve good performance even with a
training set of relatively small size, which is characteristic of parametric methods.
Moreover, in the training stage, Mixca can easily incorporate probabilistic semi-
supervision (PSS): labelling by an expert of a portion of the whole available
training set of samples. These properties of Mixca are well-suited for the particular
problem considered: classification of ceramic pieces coming from four different
periods, namely, the Bronze Age, Iberian, Roman, and the middle Ages. A set of
features is obtained from the processing of the ultrasonic signal that is recorded in
through-transmission mode using an ad hoc device. A physical explanation of the
results is obtained with comparison with classical methods used in archaeology.
The results obtained are indicative of the promising potential of ICAMM in that
particular application and in the general area of material classification [ 1 ].
The second application attempts to solve two problems in NDT of historical
building restoration using ICA: diagnosis of the material consolidation status and
determination of the thickness of the material. In those applications, the injected
ultrasonic pulse is buried in backscattering grain noise plus sinusoidal phenomena;
these phenomena are analyzed by ICA. The mixture matrix is used to extract
useful information concerning resonance phenomena of multiple reflections of the
ultrasonic pulse at non-consolidated zones and to improve the signals by detecting
interferences in ultrasonic signals. The results are shown by real experiments on a