Mazlan, Nurina Fatini (2026) Forensic discrimination of papers using atr-ftir spectroscopy and chemometrics techniques. Project Report. Universiti Sains Malaysia. (Submitted)
![[img]](http://eprints.usm.my/style/images/fileicons/application_pdf.png) |
PDF
Download (375kB) |
Abstract
Paper is a significant type of forensic document evidence, which can reveal a lot of information about the authenticity of documents and their provenience. The fact that the paper can be distinguished on the basis of brand origin can be used in the forensic investigations in the cases of document alteration, substitution, or fraud. Discrimination of modern paper materials is however difficult because they are similar in chemical composition which is mostly composed of cellulose. The research was to assess how effective Attenuated Total Reflectance fourier transform infrared (ATR-FTIR) spectroscopy in conjunction with chemometric techniques is in distinguishing between exercise book paper and various commercial brands. In the study, the exercise books which were analyzed were five brands: CAMPAP, Smart Write, KAMI, Uni Paper Products and SBS Paper Products. ATR-FTIR spectroscopy was performed on different spots of the paper samples to collect infrared spectra in order to consider surface heterogeneity and instrumental variation. Standard Normal Variate (SNV) was used to preprocess spectral data and analyze the spectral data in the fingerprint region 1500-600 cm⁻¹. The dimensionality reduction and pattern recognition were implemented using Principal Component Analysis (PCA), and the supervised classification was implemented using Linear Discriminant Analysis (LDA). The PCA output indicated that the top three principal components explained 93% of the total variance (PC1 61.6%, PC2 21.9% and PC3 9.5%), which showed that the majority of the information on spectral was well preserved. PCA score plots indicated the trends of clustering based on paper brand, but there also were a few samples partially overlapping. The next LDA model was successful in classifying the original grouped samples with 100% accuracy and cross-validation with the leave-one-out method was successful with 73.3% prediction accuracy, which is moderate discrimination ability on unknown samples. All in all, the results suggest that ATR-FTIR spectroscopy with PCA and LDA offers a non-destructive, objective and a reliable technique of discrimination and classification of paper according to brand origin. It is a powerful analytical framework that has high potential as an addition to forensic document examination.
| Item Type: | Monograph (Project Report) |
|---|---|
| Uncontrolled Keywords: | atr-ftir spectroscopy |
| Subjects: | R Medicine R Medicine > RA Public aspects of medicine > RA1001 Forensic Medicine. Medical jurisprudence. Legal medicine |
| Divisions: | Kampus Kesihatan (Health Campus) > Pusat Pengajian Sains Kesihatan (School of Health Sciences) > Monograph |
| Depositing User: | MUHAMMAD AKIF AIMAN AB SHUKOR |
| Date Deposited: | 24 Aug 2026 03:40 |
| Last Modified: | 26 Oct 2026 03:12 |
| URI: | http://eprints.usm.my/id/eprint/64826 |
Actions (login required)
 |
View Item |
