Exploratory analysis of UV-vis absorption spectra
Wally Smith, K. P. Lam · 2010 3rd International Congress on Image and Signal Processing · 2010
Independent Component Analysis (ICA) is an unsupervised approach to separating mixed independent signals. This paper investigates the potential for using ICA in the forensic analysis of black ballpoint pen inks based on UV-vis spectrophotometry data. Here, the UV-vis absorption spectra of a range of twenty-five different ballpoint pens were examined in detail, with the principal aim of determining whether two pieces of written text originated from the same pen. By adopting an ICA mixture model, the spectral data have been successfully decomposed into independent absorption patterns pertaining to the underlying chemical/physical processes. In particular, both the FastICA and JADE algorithms were able to identify and extract the compositional dye components of the inks that are either freshly prepared or have been subjected to natural light over fixed time periods, with promising results.