Raman spectroscopy imaging and chemometrics to investigate forensic documents involving delinquent actions such as crossed lines and obliteration

Abstract

Document forgery is a serious crime followed by judicial authorities. The most common cases of fraud in handwritten documents are related to character modification or text deletion, known as falsifications through crossed lines and obliterations, respectively. The use of destructive and subjective techniques in the last decades has hindered forensic investigations. This is the reason why this work proposes a non-destructive, quick, objective, and trustworthy analytical technique for forensic detection of fraud in handwritten documents, based on the combination of hyperspectral Raman imaging with the image unmixing method Multivariate Curve Resolution – Alternating Least Squares (MCR-ALS). The beginning of the research consisted of investigating the potential of Raman spectroscopy to distinguish commercial blue inks. Ink differentiation using Raman spectroscopy has proven to be very satisfactory for inks of different brands and typologies. Hence, the next step was the acquisition of Raman images in simulated cases of document falsification using confocal microscopy. The distribution maps and spectral signatures obtained for every ink in these forgeries applying spectral preprocessing and MCR-ALS have been very satisfactory, proving that the used analytical techniques are appropriate and allow obtaining detailed information regarding the distribution of inks throughout the paper in any written document. It has been observed that the nature of inks leads to different interactions in the crossed strokes and in the ink distribution on the paper. The chronological order of ink strokes application has been correctly solved in almost all crossed lines studied cases, except in a reported case of inks of a particular brand. Obliterated texts have been satisfactorily recovered, irrespective of the ink typology applied over them. We have proven that fast Raman imaging operates correctly to detect obliterations and that an increase of spatial resolution allows recovering complicated and small characters very accurately.