Speeding up macro-Raman mapping by reducing the number of sampling points

Macro-Raman mapping is an approach that allows to obtain high-resolution molecular maps of artefacts, over an area of several square centimetres. The method is based on recording thousands of spectra in a grid. The main drawback of this method is that it is very time consuming. A straightforward approach to reduce the measurement time is achieved by reducing the number of points that are measured. Not all points of the map are equally informative: Pixels that are surrounded by similar points might be less interesting to measure. Therefore, an algorithm is proposed to select where to measure and which points to omit. The missing pixels can be filled in a posteriori, by using a suitable interpolation algorithm. The selection of the omitted pixels can be performed based on information that is available from other analytical techniques. In this case, a selection was made based on the local variances in a colour picture. The approach is evaluated by recording macro-Raman maps of details of a Neptune watercolour painting on paper. In a first stage, the Raman intensities of the scaled and baseline corrected spectra at specific band positions were colour coded and plotted as Raman maps. The interpolation of the Raman maps yielded satisfying results. The image outline could clearly be identified in the maps, and the differently coloured zones were distinguished. Next to this univariate approach, it was demonstrated that also a multivariate data extraction method (principal component analysis) is compatible with the proposed algorithm to measure 25% less datapoints.