Scientists from St Petersburg University develop a new way to detect document forgery based on digital analysis of the colour of pen ink
Chemists from St Petersburg University have proposed a new approach to the classification of blue ballpoint pen ink. This method will enable forensic specialists to more accurately identify document forgery in future.
One of the major challenges in forensics is determining document forgery. Most forensic examinations to determine the age of documents are carried out using the only certified method of the Russian Ministry of Justice. It is based on the gas-liquid chromatography method that analyses the dynamics of highly volatile component evaporation. The method is highly accurate, but it is time-consuming to conduct such an examination. Moreover, only documents no older than two years can be analysed.
The findings of the research supported by a grant from the Russian Science Foundation are published in Analyst, the journal of UK’s Royal Society of Chemistry.
‘Determining the authenticity and time of creation of a document is one of the key factors in forensic technology. However, most of the solvents in the ink used to sign documents evaporate within two years, leaving only the ink base, which can be preserved for over 100 years. To increase the accuracy of examinations, we turned our attention to analysis of this component,’ explained Ilya Tumkin, Associate Professor in the Department of Laser Chemistry and Laser Materials Science, St Petersburg University.
An interdisciplinary team consisting of experts in chemistry and applied mathematics has developed a radical new approach that makes it possible to determine the authenticity and, in the long term, the age of the ink used on a document without the need for complex chemical analysis equipment. The pen ink is exposed to a special solvent that enables you to distinguish one type of ink from another. In this way, it can be reliably determined whether fragments of the text may have been falsified.
For a more accurate analysis, after using the solvent, the written text is subjected to a second processing step: scientists take a photograph of the handwritten text using monochrome LEDs in red, blue and green light separately. This makes it possible to decompose the colour of an ordinary blue pen into its components and determine, for example, which fragments of the text may have been added later, possibly by another person. The image processing to classify the ink was done using machine learning. In the long term, this technique will make it possible to identify the presence of different inks in the documents examined by forensic specialists more quickly and accurately.
The research was conducted as part of a grant from the Russian Science Foundation "Development of a method for the analysis and classification of ink compositions on paper substrates using selective extraction and digital colour analysis". The study of chemical compositions was carried out using equipment of the resource centres of the St Petersburg University Research Park: the Centre for Optical and Laser Materials Research and the Chemical Analysis and Materials Research Centre.
‘The algorithm developed on the basis of digital colour analysis has proved to be more informative than the high-performance liquid chromatography method commonly used in technical document examination. There are now well-established methods for determining only one type of dye used in writing documents − triphenylmethane dyes. However, we analyse the full composition of inks, which can include phthalocyanines dyes, triazine dyes, azo dyes and many other classes of dyes. Thus, using the developed algorithm, it is possible to obtain a more detailed clustering than before,’ said Andrei Kalinichev, Principal Investigator and Assistant Professor at St Petersburg University.