This paper concerns a first attempt of application of multivariate calibration to the field of cultural heritage for the determination of the superficial pigments composition of a painting. For this purpose, 10 mixtures of three organic pigments (Alizarin, Permanent Red and Phtalocyanine Green) selected by an augmented simplex-centroid design were mixed with linseed oil and spread on 10 cotton canvas strips. Each sample was characterised with three genuine replicates of an ATR-IR spectrum. Three calibration models, responding to the relative concentration of each specific pigment in the mixtures, were built with the partial least squares (PLS1) algorithm, using the ATR-IR spectra of the surfaces as predictors. The three mixtures corresponding to the augmentation of the simplex-centroid design were used to validate the model predictive ability which proved to be very satisfactory, with very low root mean squared error of prediction (RMSEP). Finally the models were successfully applied to a real painting, to predict the concentrations of some unknown mixtures of the three studied pigments. This work might have possible applications in the determination of the composition of dyes in real paintings to obtain information on the execution technique and for restoration purposes.
Multivariate calibration applied to the field of cultural heritage: Analysis of the pigments on the surface of a painting
MARENGO, Emilio;ROBOTTI, Elisa;
2005-01-01
Abstract
This paper concerns a first attempt of application of multivariate calibration to the field of cultural heritage for the determination of the superficial pigments composition of a painting. For this purpose, 10 mixtures of three organic pigments (Alizarin, Permanent Red and Phtalocyanine Green) selected by an augmented simplex-centroid design were mixed with linseed oil and spread on 10 cotton canvas strips. Each sample was characterised with three genuine replicates of an ATR-IR spectrum. Three calibration models, responding to the relative concentration of each specific pigment in the mixtures, were built with the partial least squares (PLS1) algorithm, using the ATR-IR spectra of the surfaces as predictors. The three mixtures corresponding to the augmentation of the simplex-centroid design were used to validate the model predictive ability which proved to be very satisfactory, with very low root mean squared error of prediction (RMSEP). Finally the models were successfully applied to a real painting, to predict the concentrations of some unknown mixtures of the three studied pigments. This work might have possible applications in the determination of the composition of dyes in real paintings to obtain information on the execution technique and for restoration purposes.File | Dimensione | Formato | |
---|---|---|---|
ACA pigments on a painting.pdf
file disponibile solo agli amministratori
Tipologia:
Altro materiale allegato
Licenza:
DRM non definito
Dimensione
514 kB
Formato
Adobe PDF
|
514 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.