Projection pursuit and PCA associated with near and middle infrared hyperspectral images to investigate forensic cases of fraudulent documents

• DOI: 10.1016/j.microc.2016.10.024
• Publication Date: 2016-10-27T04:52:04Z
• Authors: José Francielson ...; Carolina S. Silva; André Braz; Maria Fernanda Pi...; Ricardo Saldanha ...; Celio Pasquini; Peter D. Wentzell
• Keywords: Projection pursuit; White (mutation)

In forensic examination of questioned documents a type casework often encountered are the frauds that occur by mean addition and adulteration parts text or numbers on document. The goal this work is to evaluate performance hyperspectral images (HSI) in near (NIR) middle (MIR) regions, combined with unsupervised pattern recognition techniques Principal Component Analysis (PCA) Projection Pursuit (PP) for rapid, reliable non-destructive identification document falsifications means alterations additions. Blind tests were conducted purpose. Sixteen black ink pens from different brands, models types employed prepare samples two ways: (i) initial discrimination method validation, straight lines approximately 2 cm long produced white paper; (ii) blind testing, three collaborators used any sixteen available prepared genuine altered/added (in total 30 samples) paper bank check paper. Overall, PP analysis showed better results than PCA discriminate 120 pairs using HSI-MIR (97.5% 87.5%, respectively). It important mention 10.0% not discriminated PP, which highlights importance use chemometric techniques. HSI-NIR combination was able solve 76.7% 83.3% testing samples, respectively. When complementary way HSI-NIR, test increased 90%. Therefore, show great potential provide objective suspected fraudulent documents.