Fourier Transform Hyperspectral Visible Imaging and the Nondestructive Analysis of Potentially Fraudulent Documents

Abstract

The work presented in this paper details the design and performance characteristics of a new hyperspectral visible imaging technique. Rather than using optical filters or a dispersing element, this design implements Fourier transform spectroscopy to achieve spectral discrimination. One potentially powerful application of this new technology is the nondestructive analysis and authentication of written and printed documents. Document samples were prepared using red, blue, and black inks. The samples were later altered using a different ink of the same color. While the alterations are undetectable to the naked eye, the alterations involving the blue and black inks were easily detected when the spectrally resolved images were viewed. Analysis of the sample using the red inks was unsuccessful. A 2004 series $20 bill was imaged to demonstrate the application to document authentication. The results argue that counterfeit detection and quality control during printing are plausible applications of Fourier transform hyperspectral visible imaging. All of the images were subjected to fuzzy c-means cluster analysis in an effort to objectively analyze and automate image analysis. Our results show that cluster analysis can distinguish image features that have remarkably similar visible transmission spectra.