- Research Article
- Open Access
ICA Mixtures Applied to Ultrasonic Nondestructive Classification of Archaeological Ceramics
© A. Salazar and L. Vergara. 2010
- Received: 23 December 2009
- Accepted: 7 June 2010
- Published: 29 August 2010
We consider a classifier based on Independent Component Analysis Mixture Modelling (ICAMM) to model the feature joint-probability density. This classifier is applied to a challenging novel application: classification of archaeological ceramics. ICAMM gathers relevant characteristics that have general interest for material classification. It can deal with arbitrary forms of the underlying probability densities in the feature vector space as nonparametric methods can do. Mutual dependences among the features are modelled in a parametric form so that ICAMM can achieve good performance even with a training set of relatively small size, which is characteristic of parametric methods. Moreover, in the training stage, ICAMM can incorporate probabilistic semisupervision (PSS): labelling by an expert of a portion of the whole available training set of samples. These properties of ICAMM are well-suited for the problem considered: classification of ceramic pieces coming from four different periods, namely, Bronze Age, Iberian, Roman, and Middle Ages. A feature set 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 demonstrate the promising potential of ICAMM for material classification.
- Feature Vector
- Mixture Modelling
- Independent Component Analysis
- Physical Explanation
- Training Stage
To access the full article, please see PDF.
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.