Elena Rogoz*, Olimpia-Hinamatsuri Barbu**
* National University of Arts, Department of Conservation – Restoration, Bucharest, Romania; graduate student (
** Romanian National History Museum, Centre of Research and Scientific Investigation, Bucharest, Romania; researcher (
Abstract
This paper presents the advantages of using principal component analysis (PCA) and neural network (NN) to the Fourier transform infrared (FTIR) data for identifying the protein content in the painting support. We selected for PCA the region between 1500 and 1750 cm-1 of the FTIR spectra. The determination of protein concentration is based on the characteristic amido I and amido II absorption bands. In order to quantify the protein content in the painting support of two Romanian wooden churches (18th c.), we prepared fifteenconcentrations of glue mixed with the gypsum mortar. Neural network was used for predicting the protein concentration in the unknown samples, using as inputs the FTIR-PCA data of the first five components.
{flippingbook_book 33}